Firearm Handling, Use-of-Force, and Bystander Injury Litigation

• ·Purpose, and How to Use This GuideOpen
• Part 1The Case FamiliesOpen
• Part 2The Human-Factors HingeOpen
• Part 3The Standard for Justifiable ForceOpen
• Part 4Police and Civilians, ComparedOpen
• Part 5Personal Injury, Wrongful Death, and Reckless DisregardOpen
• Part 6Training, Hiring, and SupervisionOpen
• Part 7The Constitutional Standard for Police ForceOpen
• Part 8Officer-Created JeopardyOpen
• Part 9Qualified Immunity and the Municipal ClaimOpen
• Part 10State Law Above the Constitutional FloorOpen
• Part 11Excessive Force, Restraint Deaths, and the VolleyOpen
• Part 12The Bystander Injury CaseOpen
• Part 13The Scenarios and the ClaimsOpen
• Part 14Legal Theories at a GlanceOpen
• Part 15The NumbersOpen
• Part 16What the Expert Work Actually InvolvesOpen
• Part 17Evidence Preservation and the Discovery ListOpen
• Part 18Building the Opinion, Either WayOpen
• ·What Reasonable Force Looks LikeOpen
• ·Appendix: Sources and NotesOpen
• ·About the AuthorOpen

This guide is for attorneys evaluating or preparing a case in which a firearm was fired, or force was applied, by a civilian, a security guard, or a police officer against another person.

These cases split into families that get litigated, and analyzed, differently. In the first, the discharge was unintentional: a gun goes off during a struggle, a holster fails, a finger finds a trigger it shouldn't have touched. In the second, the discharge was a decision, and the question is whether that decision was justified. Treating a use-of-force case like a handling case, or the reverse, is the most common way these matters get mishandled before they ever reach a jury.

A third split runs across both. A civilian or a security guard is a private actor, judged under state tort law and state self-defense doctrine. A police officer is a state actor, and the same shooting is judged under the Fourth Amendment, filtered through qualified immunity, with a separate claim available against the department that trained him. Parts 1 through 3 apply to every shooter, badge or no badge. Parts 5 and 6 come next because that is where most of these cases live: ordinary negligence, gross negligence, reckless disregard, and the training and supervision failures that produced the shooter. None of that turns on the Constitution. Parts 7 through 12 are the constitutional track, which matters most when the defendant is a government and the claim is a civil rights claim. Part 12 covers the case where the person hit was never the target at all, which is governed by different law again.

The guide gives you the taxonomy, the standards that govern each family, the case law that controls the police side, and the legal theories that follow from each fact pattern. It runs the analysis on matters drawn from public reporting. It lists the evidence that has to be preserved before it disappears, and it says what the expert work involves. Figures come from public reporting, court records, and published litigation studies. Estimated numbers are marked as estimates.

The analytical posture

Two questions come up most often. If the discharge was unintentional, did the shooter break a firearm handling rule? If it was intended, was the decision to use force reasonable on what the shooter knew at that moment?

Plenty of cases ask something else. A bystander injury case asks whose round it was, and whether the shooter accounted for what stood behind the target. A restraint death asks how long the weight stayed on the back and when the breathing changed. A training claim asks what the agency taught and what it already knew. Getting the question right is the first job. Answering it is the second.

This guide works for either side

The same analysis that condemns one shooting exonerates another, and the method doesn't change with the retaining party. An attorney defending an officer will find four things here. The authority that bars hindsight. The case law that forecloses a bare complaint about round count. The decisions holding that a training violation isn't a constitutional violation. And the reconstruction work that shows a threat was real and immediate.

An attorney suing an officer will find three. The authority that opens the whole encounter to scrutiny. The measurements that turn a claimed split-second decision into a demonstrated interval of available time. And the theories that outlive the officer's immunity.

Part 18 puts both cases side by side. Read it first if you're triaging a matter and want to know which way the record is likely to run.

Sort the matter into the right family before anything else. The standard of care, the applicable privilege, and the theories available to each side all follow from that sort.

1. Unintentional discharge, or handling-violation casesThe shooter didn't mean to fire. A gun discharges while being drawn, holstered, or during a physical struggle over control of the weapon. The governing standard is the four cardinal rules of firearm safety, and the question is whether the shooter's handling met that standard. Foreseeability and design of the holster, retention gear, and duty belt often matter here.
2. Intended discharge, or use-of-force casesThe shooter meant to fire and claims the shooting was justified. For a private actor that means self-defense, defense of others, or in a narrow set of jurisdictions a citizen's-arrest privilege. For an officer it means the Fourth Amendment. Either way the governing standard is reasonableness: what a similarly situated, similarly trained person would have perceived and done given the totality of the circumstances at the moment force was used.
3. Hybrid casesMany real matters carry both threads. A guard draws intentionally to detain a shoplifter, meeting the threshold for a use-of-force analysis, and the gun then discharges during the ensuing struggle, meeting the threshold for a handling analysis. An officer decides to fire, which is a use-of-force question, and then fires eight rounds through glass, which raises handling, target identification, and backdrop questions as well.
4. Force short of a shotNot every excessive-force case involves a firearm. Prone restraint, neck holds, conducted-energy devices, chemical agents, impact weapons, and hard empty-hand control make up their own family. Each carries its own governing law, its own medical causation questions, and its own standard of care. Part 11 covers them.
5. The bystander injury caseThe person shot was not the person the shooter aimed at. A neighbor behind a wall, a hostage, a passerby, a clerk at a register. This family is governed by different law from every other one in this list, because a person who wasn't the target usually wasn't seized, which takes the Fourth Amendment off the table. Part 12 covers it.
6. The employer and premises layerAbove the individual shooter sits a second, often larger question: did the security company or property owner put an unqualified, undertrained, or poorly supervised armed person in that position. This layer runs on ordinary negligence principles, not on the handling or use-of-force standard that governs the shooter, and it's frequently where the largest exposure sits.
7. The public and private divideA civilian or a security guard is a private actor, judged under state tort law, self-defense doctrine, and the shopkeeper's privilege. A police officer is a state actor, and the same shooting is judged under an entirely different track: the Fourth Amendment's objective-reasonableness standard, qualified immunity, and, where a department's policy or training is implicated, municipal liability. The human-factors analysis, the four rules, and the reconstruction work are the same across both tracks. The legal theory that carries the case isn't.

Why the sort matters

A handling-violation case turns on the four rules and on equipment and training records: was the finger on the trigger, was the muzzle under control, was the holster adequate for the duty the guard was assigned. A use-of-force case turns on perception and proportionality: what did the shooter see, what alternatives existed, was the force used proportionate to the threat. Building a handling-violation record in a case that's really about the reasonableness of the decision to shoot, or the reverse, wastes discovery and weakens the eventual presentation to a jury.

The sorting questions, in order

1. Was the discharge or application of force intended? This picks the family. Video of trigger-finger position before the decision point usually answers it.
2. Was the shooter a state actor? Sworn officer, deputized guard, or private citizen. This picks the legal track and decides whether Section 1983 of Title 42 of the United States Code is available at all.
3. What was the injured person's custodial status? Free citizen, pretrial detainee, or convicted prisoner. Different constitutional amendments govern each, and the standards differ enormously in difficulty. Part 7 sets out the allocation.
4. Is there an employer or municipal layer? A private employer is reachable on respondeat superior and on direct negligence. A municipality is reachable only on policy, custom, or failure to train.
5. How many separate applications of force were there? Each one gets its own analysis. A justified first round doesn't justify an eighth.
6. Was the injured person the target? If not, the case changes tracks entirely, and the first job is proving whose round it was. Part 12 sets out how.

Strip any of these cases down and one question is left. Did the shooter handle the gun correctly and use force reasonably, judged on what he knew at that moment? That's a human-factors question. I work on how people handle guns, how they break the rules, and what governs pointing a gun at a person.

The defense opens the same way in every family

In a handling case, the defense argument is that the gun didn't malfunction, the shooter broke a rule. In a use-of-force case, the defense argument is that the force was reasonable given a genuine and immediate threat. Each defense is strongest when it's the whole story. Each falls apart the same way, when the other side shows to a professional standard that the shooter's own conduct doesn't fit that story.

The same analysis serves either side

This expertise is symmetric, and that's a strength on the stand. The method doesn't change with the retaining party. Only the facts decide the conclusion.

A neutral method built on published standards follows the evidence wherever it goes. That method survives cross-examination and an admissibility challenge. It also lets me work either side of the docket without contradiction.

The four rules, and what compliance looks like

The universal firearm safety rules are the yardstick for the handling half of any case. Each one maps to a factual question an expert can answer from the video, the physical evidence, and the witness record.

For a fuller discussion of the rules, see any of my books.

The reasonableness half: what the law asks

The four rules still govern how an intentional shot was fired. But the case turns on an earlier question. Was the shooter justified in firing at all? For a private actor that question is decided under state law of self-defense and defense of others, applied to the totality of circumstances at the moment force was used, not with the benefit of hindsight. Ask four questions. What did the shooter see? Was that reading reasonable given his training and what he knew? Did lesser options exist? Was the force he used proportionate to the threat he saw? Two facts recur across these matters and carry enormous weight: whether the subject was armed, and whether the subject was moving toward the shooter or away.

Perception, reaction, and the time the shooter actually had

Human beings don't perceive and act instantly, and the size of the gap decides cases. A person has to detect a stimulus, recognize it, select a response, and execute it. Each step takes measurable time, and the total climbs when the stimulus is ambiguous, when the choice isn't binary, when the light is poor, when the person is coming out of sleep, or when the command arrives in a language the person doesn't speak well.

Blair and colleagues, in "Reasonableness and Reaction Time," Police Quarterly (2011), put officers about ten feet from a subject with weapons already drawn and aimed, and measured what happened when the subject chose to fire. The officers generally couldn't fire before the subject did. Mean officer reaction time ran about four tenths of a second, and the subject's firing time about the same. Two conclusions follow, and they cut in opposite directions.

• For the defense. Reaction lags action. Some rounds fired after a threat ends are physiologically explainable, not evidence of malice, and an officer at contact distance genuinely has no reaction advantage.
• For the plaintiff. The same physiology means an officer who closes to contact distance has given away the only margin that would have let him respond to anything short of shooting. The research that excuses a trailing round also condemns the approach that made the round necessary.

The same arithmetic governs the person on the other end. When an officer issues a command and fires roughly a second later, the question is whether the interval allowed a human being to perceive the command, understand it, choose a response, and move. That's a measurement, not an argument, and Part 16 sets out how to take it.

Where a police-involved shooting shifts the ground

Two Supreme Court cases carry a police-involved matter that a private-actor matter never reaches. Tennessee v. Garner, 471 U.S. 1 (1985), held that shooting a fleeing suspect who presents no immediate threat is an unreasonable seizure under the Fourth Amendment. Graham v. Connor, 490 U.S. 386 (1989), set the objective-reasonableness standard courts have applied to every officer-involved force claim since. Judge the officer from the scene, not from hindsight, and allow that officers make split-second calls in tense, fast-moving conditions. That standard rests on the same human-factors reconstruction as the rest of this guide. What did the officer see, when did he see it, and what else could he have done? Qualified immunity, a defense unique to state actors, then asks a second question on top of reasonableness. Both questions benefit from the same frame-by-frame record. Parts 5 through 7 take them in order.

Four conditions have to exist together before deadly force is justified. I teach them in the Wolf Safety Series and I test every shooting against them.

These conditions govern a private shooter directly. They also state, in plain operational terms, what a police use-of-force policy asks an officer to establish before firing, so they work on both tracks. They aren't the constitutional test. Part 7 sets out that test, and Part 4 explains why the two differ.

The four conditions for deadly force

Deadly force is justified, and likely necessary, only when all four of these conditions exist together. If one is missing, the force isn't justified.

All four conditions have to be tested against the facts as they existed in that moment, not against how the situation looks afterward. A defense expert reconstructs the record to show the four conditions were met. A plaintiff's expert reconstructs the same record to show at least one of them was missing.

Deadly force and serious bodily injury, defined

Deadly force is any type or degree of force that's likely to result in death or serious bodily injury. Serious bodily injury covers injuries that cause death, serious permanent disfigurement, or long-term loss or impairment of the function of any body part. A firearm isn't the only means of deadly force. A fist, a vehicle, a blunt object, a neck restraint, or sustained pressure on a prone person's back can all meet the definition. It depends on how mismatched the two people are and how long the pressure lasts.

Factors that weigh on the reasonableness of the belief

1. Number of assailantsThree attackers against one defender changes the ability and opportunity analysis even if no single attacker is armed.
2. Physical disparitySize, strength, age, number, and known skill can turn empty hands into a means of deadly force.
3. Weapons, visible or impliedA weapon doesn't have to be seen to be reasonably believed present. A hand concealed in a waistband during a verbal threat can support the same belief a visible weapon would. The converse also holds, and it's where a great many shootings go wrong. An object that looks like a gun isn't one. Whether the mistake was reasonable is a separate question from whether the tactics that forced a snap call were reasonable.
4. Distance and closure rateDistance is time, and time is the only thing that lets a person verify a threat before responding to it. Closure rate matters more than static distance.
5. Cover and barriersA wall, a vehicle, a door, or a fence between the parties changes the ability and imminence analysis. So does the decision to give up cover that was available.
6. Lighting and visibilityNight, backlighting, a flashlight aimed at the eyes, tinted or wet glass, and rain all degrade identification. They degrade it for both parties, which is why they cut both ways.
7. Condition of the subjectIntoxication, mental health crisis, deafness, limited English, and disorientation on waking all change how a person will respond to a command. If the shooter knew or should have known about the condition, judge him on it.

The force continuum

The continuum is the training model that sits underneath the reasonableness standard. It lays out the responses available before force ever reaches the four conditions above. Most security and law enforcement programs teach from it.

The same shooting, by two different shooters, produces two different cases. Not two different outcomes on the same test, two different tests, with different elements, different burdens, different defenses, and different people paying the judgment.

An attorney who understands only one track will misplead the other.

The structural difference: a defense versus a claim

A civilian who shoots someone raises a justification defense inside somebody else's case. The prosecutor charges assault or homicide, or the estate sues for battery, and the shooter answers that the act was privileged. The question is whether the shooter's belief excuses conduct that would otherwise be a crime or a tort.

A police officer who shoots someone faces a claim that the government seized a person unreasonably. The shooting is a Fourth Amendment event from the moment the round strikes, and the question is whether that government intrusion was reasonable. Nobody asks what the officer believed. Graham v. Connor removed intent from both sides of the analysis: an officer's bad intentions won't make a Fourth Amendment violation out of objectively reasonable force, and his good intentions won't make objectively unreasonable force constitutional.

The officer is judged by a standard that ignores what he was thinking and forgives what he couldn't have known, but holds him to what a trained professional should have done. The civilian is judged by a standard that examines what he actually believed and why, but forgives him for having no training at all.

The arrest privilege, and why it decides so much

An officer with probable cause may seize a person, may use objectively reasonable force to make the arrest, may use force to overcome resistance, and may continue until the arrest is complete. California Penal Code section 835a(b) states it plainly, and most states codify the same authority.

That authority changes everything downstream. The officer may lawfully start the confrontation. When the subject resists a lawful arrest, the resistance itself justifies escalation. Walking up in the first place isn't normally actionable.

A private person has no general power to seize anyone. Three narrow doctrines fill that gap, and each one is a trap for a shooter who exceeds it.

• Citizen's arrest. Statutory, narrow, and shrinking. Georgia repealed its 1863 statute outright in 2021 by House Bill 479, after the killing of Ahmaud Arbery. What replaced it runs only to retail and food-service operators, business owners, weight inspectors, and licensed security personnel. Deadly force is excluded except in self-defense, defense of habitation, or prevention of a forcible felony.
• Shopkeeper's privilege. Recognized in substantially every state, tracking the Restatement (Second) of Torts section 120A. Detention on reasonable grounds, for a reasonable time, in a reasonable manner, using reasonable non-deadly force only. It isn't a defense to a shooting, because by its own terms it excludes deadly force.
• Defense of others. Not a seizure power at all. It privileges force to prevent harm to a third person, and in most states the defender stands in the shoes of the person defended.

In an officer case, starting the encounter is presumed lawful and the plaintiff has to attack the force used inside it. In a civilian case, how the shooter started it is frequently the whole fight. If the civilian had no privilege to detain, then every touching is a battery, the person detained had a right to resist, and the civilian is the aggressor. An aggressor forfeits self-defense in every state unless he withdraws and communicates the withdrawal. The civilian who chases and confronts isn't a defender. He's the initial aggressor, and his later claim of fear is legally beside the point. That's the theory on which the Arbery defendants were convicted.

Retreat, and preclusion

An officer making or attempting an arrest has no duty to retreat and doesn't become an aggressor by using objectively reasonable force. California Penal Code section 835a(d) and Texas Penal Code section 9.51(e) are representative. The rule follows from the privilege: an officer with a duty to complete an arrest can't simultaneously have a duty to abandon it.

For civilians, three regimes operate, with the castle exception overlaying all of them. Roughly two thirds of the states are stand-your-ground, removing the retreat duty anywhere the actor has a lawful right to be, provided he didn't provoke and, in some states, isn't engaged in criminal activity. The rest impose a duty to retreat outside the home when the actor knows he can get away in complete safety. What he knew is subjective, so the jury decides it. Castle doctrine removes the retreat duty in the dwelling, and in many states in an occupied vehicle and place of business. Some states add a presumption that fear was reasonable against a forcible intruder, and Colorado goes further, granting immunity from criminal prosecution and civil liability for force against an intruder in a dwelling.

Preclusion, the fourth condition in Part 3, has no officer counterpart at common law. An officer making a lawful arrest never has to try walking away first. Several states have now imposed something close to preclusion by statute, requiring officers to exhaust reasonable alternatives before deadly force. That's a statutory duty, not a constitutional one, and Part 10 identifies where it applies.

The fleeing suspect

An officer may use deadly force against a fleeing suspect where there's probable cause to believe the suspect poses a significant threat of death or serious physical injury to the officer or others, and where feasible, after a warning. That's Tennessee v. Garner, and it's a threat-based rule, not an offense-based one.

A civilian generally has no fleeing-felon privilege at all. The modern codes are close to uniform, and several are explicit that the arrest deadly-force privilege runs only to officers or to a person acting at an officer's direction. Mississippi Code section 97-3-15 and Texas Penal Code section 9.51(d) and (g) both read that way. Texas surprises people here, because its reputation invites the opposite assumption. Anyone other than a peace officer may use deadly force to make an arrest only in an officer's presence and at his direction. Section 9.51(g) allows nothing wider.

New York Penal Law section 35.30(4) is the one clean statutory survivor. It lets a private person acting on his own account use deadly physical force to arrest someone in immediate flight from murder, first-degree manslaughter, robbery, forcible rape, or forcible criminal sexual act. Note the trap in the text: the offense has to have actually been committed, not merely reasonably believed. Repeal legislation has been introduced repeatedly. Confirm the current status before relying on it.

California Penal Code section 197(4) still reads broadly on its face, permitting justifiable homicide when necessarily committed in attempting by lawful means to apprehend a person for a felony. People v. Ceballos, 12 Cal. 3d 470 (1974), narrowed the felony-based justification to forcible and atrocious crimes threatening death or great bodily harm, and Assembly Bill 392 in 2019 rewrote the officer authority without amending section 197(4). Whether a private person can still invoke it for a fleeing felon is unsettled.

Defense of property, stated precisely

Deadly force isn't privileged to protect property alone. Restatement (Second) of Torts sections 79 and 85 state the rule. Katko v. Briney, 183 N.W.2d 657 (Iowa 1971), applied it to a spring gun in an unoccupied farmhouse and affirmed compensatory and punitive damages against the property owner. People v. Ceballos applied it to a trap gun in a garage. Both rest on the same ground. Human life outweighs property, and a machine can't judge what a person can.

Most states reach property-protective outcomes through a different door: the enumerated forcible-felony provision inside the defense-of-person statute. Texas Penal Code section 9.32(a)(2)(B) permits deadly force to prevent aggravated kidnapping, murder, sexual assault, aggravated sexual assault, robbery, or aggravated robbery. Robbery and burglary of an occupied dwelling are property-adjacent crimes that carry a threat to persons, which is why the privilege attaches. The defense is still a defense of persons.

Texas is the one state with a freestanding deadly-force privilege for property, at Penal Code section 9.42, and the provision is far narrower than it reads. Three requirements are conjunctive. The actor has to be justified in using ordinary force under section 9.41 first. The force has to be immediately necessary to prevent arson, burglary, robbery, aggravated robbery, theft during the nighttime, or criminal mischief during the nighttime, or to prevent a person fleeing immediately after burglary, robbery, aggravated robbery, or theft during the nighttime from escaping with the property. And the actor has to reasonably believe the property can't be protected or recovered by any other means, or that non-deadly force would expose someone to a substantial risk of death or serious bodily injury.

That last requirement is the one that swallows the privilege. Calling the police is another means. Photographing a license plate is another means. Filing an insurance claim is another means. In practice section 9.42 comes close to demanding impossibility. Every operative clause turns on "reasonably believes," so a jury judges the actor's belief and defines reasonableness after the fact.

Immunities, and who actually pays

• Section 1983 reaches only state actors. Title 42 of the United States Code, section 1983, has no application to a private citizen who shoots another private citizen. Bivens v. Six Unknown Named Agents, 403 U.S. 388 (1971), is the only route to a federal officer. Egbert v. Boule, 596 U.S. 482 (2022), and Goldey v. Fields, decided June 30, 2025, have cut it to almost nothing outside the original context. In a matter involving federal personnel, the constitutional damages remedy is largely unavailable regardless of the force used, and the case moves to the Federal Tort Claims Act or to state-law claims.
• Qualified immunity is available only to individual state actors. It has no application to a private shooter, and none to a municipality. Part 9 covers it.
• Municipalities have no qualified immunity and pay no punitive damages. Owen v. City of Independence, 445 U.S. 622 (1980), and City of Newport v. Fact Concerts, 453 U.S. 247 (1981). That combination is why the municipal claim is often the only route to full-value recovery when the officer wins immunity, and why the municipal defendant fights it hardest.
• State tort claims acts are the obstacle most often missed. Notice-of-claim deadlines are short, jurisdictional, and unforgiving, running from ninety days in several states to a year in others. Damage caps apply in many states. Official immunity protects discretionary acts in most. None of these apply to a Section 1983 claim, which is a reason to plead federally even in a strong state-law case.
• Respondeat superior works against a private employer and not against a municipality. That asymmetry frequently makes the private security defendant the better target, and it's why a contract guard case can be worth more than the same shooting by a sworn officer.

On who pays: Joanna Schwartz, "Police Indemnification," 89 New York University Law Review 885 (2014), studied 81 law enforcement agencies across 2006 to 2011 and found that governments paid 99.98 percent of the roughly $730 million recovered in damages. Officers essentially never contributed personally, including when they were disciplined, terminated, or criminally prosecuted. The exception proves the rule. A Dallas jury returned roughly $98.65 million against an off-duty officer personally in November 2024, and nobody will collect it.

Training, and what each side is measured against

Graham asks what a reasonable officer on the scene would do. That officer isn't a reasonable person. He's someone who completed a certified academy, holds a state certification, and works under a written policy. Every one of those is discoverable and every one supplies a yardstick.

The Bureau of Justice Statistics survey of state and local law enforcement training academies for 2022 reported that basic training averaged 806 hours overall, with firearms skills averaging 73 hours, defensive tactics 64 hours, de-escalation and verbal skills 22 hours, and response to mental illness and behavioral health 21 hours. Seventy-five percent of recruits received use-of-force simulator training, ninety-eight percent trained under simulated stress, and ninety-six percent received training on identifying or responding to another officer's excessive force. Eighty-nine percent of academies required field training after the basic course.

The armed private security guard carries the same weapon under a fraction of that. Several states require no individual state license at all and regulate only the employer. Where states do license, mandated firearms training commonly runs between eight and forty-seven hours, and unarmed training between eight and forty-two. Some states leave licensing to cities. Set that against 73 academy hours plus recurring qualification plus field training plus in-service, and the disparity is the point: the guard carries the same weapon with a fraction of the training and none of the legal authority.

The asymmetry, and how it cuts

The mirror image holds for the private shooter. No qualified immunity, no arrest privilege to justify starting the encounter, no indemnification unless an employer or a homeowner's policy supplies it. But also a defense with a subjective component the state has to disprove beyond a reasonable doubt in the criminal case, no retreat duty in most states, castle protections, and in some states a pretrial immunity hearing that ends the matter before a jury sees it. The civilian's exposure is more personal and less institutional. The officer's is more institutional and harder to reach.

Most practitioners reach first for the Fourth Amendment. It's the hardest theory in the box, and in some states it's the only one that works. In others it's the one to plead last.

A shooting is also a personal injury case, a wrongful death case, a negligent training case, and sometimes a crime. Each of those runs on its own elements, its own burden, its own immunities, and its own damages. Plead them together, because they fail separately. Part 6 takes the training claim on its own, because it's usually the strongest one in the file.

What each theory actually asks

Read the second row against the first. The negligence claim asks about care, not reasonableness under the Constitution, and no qualified immunity answers it. The cost is that the recovery is often capped and there's no fee shifting. That trade is the whole strategic question in this Part.

The negligence case, and the standard that governs it

The officer isn't measured against a reasonable person. He's measured against a reasonable officer with similar skill, training, and experience. Bassett v. Lamantia, 2018 MT 119, 391 Mont. 309, 417 P.3d 299 (2018), states it in those words. That's a professional negligence standard, and every hour of training the officer received becomes a specification he can be held to.

Two decisions carry the theory further than the Fourth Amendment reaches.

Six sources supply the standard of care, in roughly descending order of weight. A statute fixing a use-of-force standard. The agency's own written policy. The training the officer actually received, which is the agency's own evidence of what it thought reasonable care required. State peace officer standards and training requirements. National model policies and generally accepted practices, which come in through expert testimony as evidence of custom. And for a guard, the state licensing statute.

Negligence per se, and where it works

A statutory violation establishes breach where the plaintiff is in the class the statute protects and the injury is the type it was meant to prevent. States split three ways on the effect. Most treat an unexcused violation as conclusive. About a dozen treat it as some evidence for the jury. A handful treat it as a rebuttable presumption.

Pick the right statute. A penal use-of-force provision is a weak candidate, because it tells an officer when force is permitted rather than what care to use, and courts decline the doctrine where a statute confers discretion or states a justification. Training and licensing statutes are the strong candidates. A minimum-hours regulation, or an armed-guard statute requiring a firearms course before a guard may carry, is prescriptive, protects the public exposed to armed personnel, and targets the exact harm.

The public duty doctrine, and the answer to it

A duty owed to the public at large is owed to no one in particular. That rule, traced to South v. Maryland, 59 U.S. 396 (1855), is the first thing a defendant raises against a state-law claim, and it kills failure-to-protect cases.

It usually has no application to a shooting, and the reason is the distinction between misfeasance and nonfeasance. Misfeasance is active misconduct working positive injury. Nonfeasance is failing to protect from a harm the officer didn't create. The doctrine addresses the second.

New York went the other way, and a practitioner who states the misfeasance rule as universal will be corrected. Ferreira v. City of Binghamton, 38 N.Y.3d 298 (2022), holds that the special duty requirement applies to every negligence claim against a municipality acting in a governmental capacity, including one where the municipal employee inflicted the injury himself. Police shot an unarmed man asleep on a couch during a no-knock entry, and the plaintiff still had to prove special duty.

He proved it. New York recognizes three routes to special duty, and the third one won: the municipality took positive control of a known and dangerous safety condition. Officers executing a no-knock warrant "take control of the targeted premises, knowingly creating an unpredictable and potentially dangerous condition." Whether that reaches a street encounter or a traffic stop is open.

Keep two New York doctrines apart, because practitioners conflate them constantly. Special duty is an element of the plaintiff's claim, pleaded and proved by the plaintiff, resolved first. The professional judgment rule is an affirmative defense, pleaded and proved by the municipality, reached only if special duty is established. And the professional judgment immunity vanishes where the municipality violates its own internal rules, which is the plaintiff's opening.

State immunity, and the honest counterweight

Courts characterize the decision to shoot as discretionary almost everywhere, and discretionary immunity is the central obstacle to the state-law claim. Cameron v. Lang, 274 Ga. 122, 549 S.E.2d 341 (2001), says it outright: firing a gun at a suspect is a discretionary act.

Three ways around it.

1. Attack the pre-shooting conduct instead. Tactical decisions get characterized as operational rather than policy-level, and they're frequently governed by mandatory training directives. Reframe from "the decision to shoot" to "the decisions that made shooting necessary." This is why Hayes and Beltran-Serrano matter so much.
2. Find the mandatory provision. Immunity doesn't protect the violation of a directive that removes discretion. "Shall not discharge when it will unnecessarily endanger innocent persons" confers judgment. "Shall not shoot at or from a moving vehicle" doesn't. Build the discovery plan around finding language of the second kind.
3. Plead the intentional tort. Discretionary immunity in most formulations doesn't extend to malice, willful and wanton conduct, or intentional injury.

Respondeat superior, which is the largest structural advantage

Under state tort law a municipality generally is vicariously liable for an officer's torts within the scope of employment. No policy. No custom. No final policymaker. No deliberate indifference. No pattern of prior incidents.

That's the single biggest difference between the state claim and a municipal claim under Monell, and it's the reason to plead state law even where the federal claim looks strong. California Government Code section 815.2(a) is the statutory illustration, and Mary M. v. City of Los Angeles, 54 Cal. 3d 202 (1991), holds that respondeat superior applies to public and private employers alike.

Six things cut against it, and all six need checking in the forum.

• Derivative immunity. California Government Code section 815.2(b) and Illinois 745 Illinois Compiled Statutes 10/2-109 both provide that where the employee is immune, the entity is immune. In a strong official-immunity state that erases the advantage.
• Intentional-tort exclusions in the tort claims act, which is the Texas problem above.
• Scope-of-employment fights. Some states put intentional torts categorically outside the scope. California holds that an officer's abuse of authority can be within it precisely because the job supplied the authority.
• The Florida inversion. Under Florida Statutes section 768.28(9)(a) the entity is the exclusive defendant unless the officer acted in bad faith or with malicious purpose or wanton and willful disregard. Prove that standard and liability moves off the entity and onto an individual who probably can't satisfy a judgment. That's a genuine fork, and you have to choose.
• No punitive damages against a public entity, nearly everywhere.
• Caps. Roughly half of municipal governments accept respondeat superior liability for this conduct under state law, and a majority of those states cap the award, commonly between $100,000 and $500,000.

State the trade plainly rather than selling one side of it. State tort law gives respondeat superior, a lower liability standard, and no qualified immunity. It then caps the recovery, denies punitives against the entity, and imposes a notice deadline as short as ninety days. Section 1983 carries fee shifting under section 1988(b) with no state equivalent in most jurisdictions, which for many cases is the decisive economic fact. Plead both.

Wrongful death and survival are two claims with two owners

The distribution consequence gets overlooked. Where a decedent carried substantial debt, the survival recovery can be consumed by creditors while the wrongful death recovery passes to the family. Plead and try them as separate claims with separate verdict lines.

Most of the survival value in a shooting case sits in conscious pain and suffering before death, and that's where the defense fights hardest, arguing instantaneous death or immediate unconsciousness. The proof is medical examiner testimony, scene audio, body-worn camera timestamps, and bystander video establishing an interval of awareness.

Two states are extreme outliers on wrongful death damages. New York allows pecuniary injury only, so grief, anguish, and loss of companionship aren't recoverable, under a statute dating to 1847. The Grieving Families Act that would change it has been vetoed four times, most recently in December 2025, and isn't law. Alabama permits only punitive damages, not compensatory. The other forty-eight allow some noneconomic recovery, and several expanded it recently.

The deadline is the trap, not the limitations period. Notice of claim runs ninety days in New York and New Jersey, six months in California and Texas, one hundred eighty days in Arizona, Idaho, and Utah. A New York municipal wrongful death case is functionally a ninety-day case. Calendar it the day you're retained.

The family's own claims

Negligent infliction of emotional distress runs on one of two rules, and the forum's choice decides whether a parent who watched has a claim.

Under the zone of danger rule the plaintiff has to have been within the physical danger the defendant created and to have feared for his own safety. In a shooting that means within the field of fire, which isn't fanciful: stray rounds, overpenetration, and crossfire regularly put family members there. In a zone-of-danger state, develop the ballistic evidence to place the parent in the line of fire.

The bystander rule comes from Dillon v. Legg, 68 Cal. 2d 728 (1968), and Thing v. La Chusa, 48 Cal. 3d 644 (1989), tightened its factors into elements. The plaintiff has to be closely related to the victim. He has to be present at the scene when the injury happens and aware at that moment that it's causing injury. And he has to suffer serious emotional distress as a result. Parent and child satisfies the relationship. The fight is over presence and contemporaneous awareness. A parent watching from a doorway generally wins. A parent who heard the shots and came out generally loses.

The relational approach is the modern trend and the position of Restatement (Third) of Torts section 48. A substantial minority keeps the zone of danger rule, and a few states still require physical impact. Don't publish a count. Check the forum.

Two other routes are underused. Restatement (Second) of Torts section 46(2) allows a member of the victim's immediate family who is present at the time to recover for intentional infliction, with no zone-of-danger requirement, no physical manifestation requirement, and no awareness element. That's materially easier than negligent infliction, and it should be pleaded alongside it where the shooting was intentional and unjustified. And note that the strongest facts for intentional infliction in these cases are usually post-shooting: delayed or denied aid, taunting, a body left uncovered in public view, conduct toward the family at the scene.

Loss of consortium follows an odd pattern worth knowing. Spousal consortium is recognized nearly everywhere. A child's claim for injury to a parent, and a parent's claim for injury to a child, are minority rules. But the corresponding death claims are near-universal, because wrongful death statutes independently authorize loss-of-society damages in forty-eight states. So in some states a family recovers more when the victim dies than when the victim survives, and counsel shouldn't assume the surviving-victim case is the easier one on damages.

The culpability ladder, and what each rung unlocks

Moving up the ladder changes what's available, and the words are used loosely enough that they have to be pinned to the forum's definition.

1. Ordinary negligence. Failure to use reasonable care. Objective, no awareness of risk required. Compensatory damages only.
2. Gross negligence. In most states not a separate mental state but negligence of an aggravated degree. California calls it "a want of even scant care or an extreme departure from the ordinary standard of conduct." Texas requires by statute both an objectively extreme degree of risk and the actor's actual subjective awareness of it.
3. Recklessness, willful and wanton conduct, conscious disregard. Restatement (Second) of Torts section 500 defines reckless disregard as acting, or failing to act where there's a duty, knowing or having reason to know of facts that would lead a reasonable person to realize not only that the conduct creates an unreasonable risk of physical harm but "that such risk is substantially greater than that which is necessary to make his conduct negligent." Restatement (Third) section 2 tightens the knowledge element and adds a burden-of-precaution test, which is friendlier to a plaintiff's expert because it invites proof that the omitted precaution was cheap relative to the risk.
4. Intentional conduct and actual malice. Purpose to cause the consequence, or knowledge that it's substantially certain to follow.

Most states treat recklessness, conscious disregard, and reckless indifference as the same standard. Ohio doesn't, and Anderson v. Massillon, 134 Ohio St. 3d 380 (2012), holds the terms describe different degrees of care and aren't interchangeable. Note Ohio's ordering, which surprises people: wanton misconduct, the failure to exercise any care where there's a great probability of harm, is more culpable than reckless conduct, the conscious disregard of a known or obvious risk. In an Ohio case the immunity exception is keyed to those words, so the distinction decides outcomes.

Illinois runs the other way and treats willful and wanton as a sliding category. Ziarko v. Soo Line Railroad, 161 Ill. 2d 267 (1994), calls it "a hybrid between acts considered negligent and behavior found to be intentionally tortious," which "may be only degrees more than ordinary negligence" in one case and "only degrees less than intentional wrongdoing" in another.

Conscious disregard for human life

This is the phrase that carries punitive damages, pierces immunity in several states, and has a criminal twin. It's also the finding an expert is most often asked to support without being allowed to state it.

On the civil side, California Civil Code section 3294 is representative: punitive damages require despicable conduct carried on with a willful and conscious disregard of the rights or safety of others. Minnesota requires clear and convincing evidence of a deliberate disregard for the rights or safety of others, which exists where the defendant knew of facts creating a high probability of injury and deliberately proceeded in conscious disregard of it.

On the criminal side the analogue is depraved heart murder, or extreme indifference to the value of human life. Note one important limit: State v. Noor, 964 N.W.2d 424 (Minn. 2021), held that depraved-mind murder requires a generalized indifference to human life and can't be committed against a particular person, which vacated a conviction in an officer-involved shooting. Not every state agrees, but the reasoning travels, and it's why depraved-heart charges are a poor fit for a shooting aimed at someone.

Punitive damages

• Against an individual officer under Section 1983, Smith v. Wade, 461 U.S. 30 (1983), permits punitives where the conduct is "motivated by evil motive or intent, or when it involves reckless or callous indifference to the federally protected rights of others." The test is disjunctive. Recklessness alone suffices, and no evil motive has to be shown.
• Not against a municipality. City of Newport v. Fact Concerts, 453 U.S. 247 (1981). The same bar reaches official-capacity claims, which are claims against the entity, so sue the officer individually if punitives matter.
• Not against the United States. 28 United States Code section 2674.
• Against a private company, most states apply the complicity rule of Restatement (Second) of Torts section 909, so the plaintiff has to reach the company's own conduct. Subsection (b), that the agent was unfit and the principal was reckless in employing him, is the negligent-hiring route to punitives, and it's the reason the personnel file matters so much.
• Constitutional limits. BMW of North America v. Gore, 517 U.S. 559 (1996), and State Farm v. Campbell, 538 U.S. 408 (2003), set the guideposts: reprehensibility, ratio, and comparable sanctions. "Few awards exceeding a single-digit ratio between punitive and compensatory damages, to a significant degree, will satisfy due process." Two of the five reprehensibility factors are physical harm and reckless disregard of health or safety, and a shooting hits both at maximum, which supports a ratio at the high end. Expect the defense to quote the qualifier: where compensatory damages are substantial, a ratio closer to one to one may be the outer limit.
• State caps don't apply to Section 1983 punitives, which are governed by federal common law. Felder v. Casey, 487 U.S. 131 (1988), supplies the preemption reasoning.

Criminal exposure, and what it does and doesn't decide

The ladder runs from murder through the endangerment offenses, and the rung that matters most in this field is the one people forget.

• Reckless or wanton endangerment reaches conduct that hurt nobody. Kentucky Revised Statutes section 508.060 makes it first-degree wanton endangerment to wantonly engage in conduct creating a substantial danger of death or serious physical injury "under circumstances manifesting extreme indifference to the value of human life," and a recent amendment elevates it where the person discharges a firearm. New York Penal Law section 120.25 requires conduct creating a grave risk of death "under circumstances evincing a depraved indifference to human life." First-degree reckless endangerment in New York is depraved-indifference murder minus the death.
• Manslaughter and criminally negligent homicide are where officer prosecutions usually land when they land at all.
• Section 242 of Title 18 reaches deprivation of rights under color of law, and its obstacle is willfulness. Screws v. United States, 325 U.S. 91 (1945), requires specific intent, though not knowledge of the constitutional provision by name. Prosecutors prove it the way civil plaintiffs prove conscious disregard: training on the limits, the written policy embodying them, prior counseling, and conduct departing from both.
• Official misconduct and assault with a deadly weapon round out the charging options.

An acquittal has no preclusive effect in the civil case, because the burdens differ. The plaintiff may still prove the same facts by a preponderance, and in most courts the acquittal isn't even admissible for its truth. A conviction frequently cuts the other way and is preclusive against the defendant, because the criminal jury found the facts beyond a reasonable doubt.

Choosing the theory

The shooter had a decision window measured in fractions of a second. The agency that hired him, armed him, and trained him had years.

That asymmetry is why the training claim is often the strongest one in the file. It also survives things that kill the claim against the individual. Qualified immunity doesn't reach a municipality. A criminal acquittal says nothing about whether the department taught what it should have. And an officer who wins on reasonableness can still have been set up to fail by an employer who never taught him the thing that would have prevented the shooting.

Two entirely different claims go by the name "failure to train," and confusing them costs cases. One runs on ordinary negligence. The other runs on deliberate indifference and usually needs a documented pattern of prior incidents. Take them in that order, because the easier one is the one most practitioners skip.

Two versions of the same claim

A state negligent training claim asks whether a reasonable employer would have trained differently and whether the failure caused the shooting. That's an expert testimony question. A Monell training claim asks for a documented pattern of similar prior incidents a policymaker knew of and ignored. That's an archival question, and the archive usually doesn't exist.

The elements are an employment relationship, an employee who was unfit or dangerous, an employer who knew or should have known, causation, and damages. Some states require that the act occurred on the employer's premises or with the employer's chattel, which an agency-issued firearm satisfies without argument.

Against a public entity the theory often dies on immunity. California bars direct entity liability for negligent hiring and supervision absent a specific statute, under Government Code section 815 and De Villers v. County of San Diego. The workaround is to sue for the supervisor's negligence as an employee tort the entity answers for vicariously, which then runs into derivative immunity. Against a private security company none of that applies.

Negligent entrustment of a firearm

An employer who puts a weapon in the hands of someone unfit to carry it is directly liable under Restatement (Second) of Torts section 390. The elements are entrustment of the chattel, an entrustee unfit to use it safely, the defendant's knowledge or reason to know of the unfitness, negligent or dangerous use, causation, and damages.

Note the knowledge standard. Courts predominantly apply section 390's "knows or has reason to know," not the broader "should know" of section 308. That's the difference between "the disqualifying record was in the personnel file" and "a careful employer would have looked."

The theory earns its place because it's direct, not vicarious. It doesn't depend on scope of employment, so it reaches the employer who armed an unfit guard even where the shooting was off duty, outside the scope, or intentional.

The constitutional version, and the footnote that matters

City of Canton v. Harris, 489 U.S. 378 (1989), holds that inadequate training supports municipal liability only where the failure amounts to deliberate indifference to the rights of persons with whom the police come into contact, and only where it reflects a deliberate or conscious choice by the municipality. The identified deficiency has to be closely related to the injury and has to have actually caused the violation. Later cases require the policy to be the moving force.

Footnote 10 of Canton is the most useful sentence in this canon for a use-of-force expert. City policymakers know to a moral certainty that officers will have to arrest fleeing felons and that officers are armed. Training officers on the constitutional limits of deadly force is so obviously necessary that skipping it can amount to deliberate indifference with no prior pattern at all. That footnote is the origin of single-incident liability, and it's written about deadly force training specifically.

Connick v. Thompson, 563 U.S. 51 (2011), then held that a pattern of similar violations is ordinarily necessary to show deliberate indifference. Read the distinction carefully before conceding it. The Court declined to apply the Canton hypothetical to prosecutors because attorneys already have professional legal training and ethical obligations, so a discovery violation isn't the obvious consequence of a lack of in-house training. That reasoning doesn't transfer to police recruits, and the Court didn't disturb footnote 10 as applied to armed officers and deadly force.

Board of County Commissioners of Bryan County v. Brown, 520 U.S. 397 (1997), sets the negligent-hiring bar: a single hiring decision supports liability only where adequate scrutiny of the applicant's background would lead a reasonable policymaker to conclude that the plainly obvious consequence would be a deprivation of rights. Rigorous standards of culpability and causation apply.

What proves it

A police-practices expert does his most consequential work here, because this count outlives the officer's immunity. The five building blocks below serve the state claim and the constitutional one alike. What changes is how much of it you need.

• Policy analysis. The agency's written use-of-force, vehicle-pursuit, shooting-at-vehicles, prone-restraint, crisis-intervention, and reporting policies, compared against the constitutional floor and against recognized benchmarks, and against the agency's own prior policy versions. Two findings pay. One is a policy that sits below the constitutional floor on its face. The other is a policy that reads well and that the agency contradicts in practice.
• Training records. Academy and in-service curricula, lesson plans, hours, qualification records, simulator logs, remedial training, and the individual officer's training file. The question isn't whether there was training. It's what was taught, how often, on what scenarios, and whether it matched the written policy. Lombardo is the model: the Supreme Court itself pointed to the department's own training warning about prone-restraint suffocation.
• Prior incident and complaint data. Use-of-force reports, citizen complaints, internal affairs files, early-intervention system data, prior lawsuits and settlements, and any monitor findings. Factual similarity is the battleground. A pattern of excessive-force complaints is nearly worthless. A pattern of the same maneuver, in the same unit, in the same circumstances, previously reported and not corrected, is the case. Three strong comparators beat ten weak ones.
• Supervisory and investigative review. How the agency investigated prior incidents. Perfunctory investigations that always end in a within-policy finding are the strongest ratification and custom evidence available. That pattern shows notice and a decision not to act.
• Causation. You have to connect the identified deficiency to this officer's specific decision in this incident. "Better training would have helped" fails. "This officer was never trained on X, the agency knew officers encountered X routinely, officers had previously mishandled X on a documented number of occasions, and this officer did precisely the thing the missing training would have addressed" is a municipal opinion.

The four attacks, and the honest answers

Decertification, licensing, and the record nobody keeps

Forty-nine peace officer standards and training agencies, the states plus the District of Columbia, contribute to the National Decertification Index, which held over 53,500 records as of its 2024 whitepaper. Reporting is voluntary in many states, which is the gap that makes the wandering-officer problem possible.

The federal counterpart is gone. The National Law Enforcement Accountability Database launched in December 2023, held 4,790 qualifying incidents across ninety-four federal agencies, and was decommissioned in January 2025. It covered federal officers only.

Private security has no equivalent at all. Nearly every state licenses both the company and the individual guard, with a separate firearms endorsement for armed guards, usually through a consumer affairs or state police agency rather than a training commission. There's no index.

These records matter to the civil case in four ways. A decertification or revocation is a finding by a neutral body. The application and the background check that was or wasn't run are the negligent hiring case. A record the employer could have found and didn't is the notice evidence. And the absence of any national record is itself the argument that a careful employer had to look harder, not less hard.

What to demand, and what it proves

The training case is built from documents, and the documents exist. Agencies and security companies generate them because licensing bodies, accreditors, and insurers require them.

Six Supreme Court decisions do most of the work in a police use-of-force case. Learn them in order, because each one narrows or widens the frame the next one operates inside.

Graham v. Connor is the frame for everything

Graham v. Connor, 490 U.S. 386 (1989), decided unanimously, holds that every claim of excessive force during an arrest, an investigatory stop, or any other seizure of a free citizen is analyzed under the Fourth Amendment's objective-reasonableness standard, not under substantive due process. Dethorne Graham was a diabetic in insulin reaction who hurried out of a convenience store, and the stop that followed left him with a broken foot, cut wrists, a bruised forehead, and an injured shoulder.

The Court listed three factors, and they're factors rather than elements:

1. the severity of the crime at issue;
2. whether the subject poses an immediate threat to the safety of the officers or others;
3. whether the subject is actively resisting arrest or attempting to evade arrest by flight.

Four sentences from the opinion get quoted in every one of these cases, and an expert should know them cold. Reasonableness "must be judged from the perspective of a reasonable officer on the scene, rather than with the 20/20 vision of hindsight." The calculus "must embody allowance for the fact that police officers are often forced to make split-second judgments, in circumstances that are tense, uncertain, and rapidly evolving, about the amount of force that is necessary in a particular situation." The question is whether the actions are objectively reasonable "without regard to their underlying intent or motivation." And: "An officer's evil intentions will not make a Fourth Amendment violation out of an objectively reasonable use of force; nor will an officer's good intentions make an objectively unreasonable use of force constitutional."

Tennessee v. Garner: the threat, not the offense

Tennessee v. Garner, 471 U.S. 1 (1985), held that deadly force to prevent the escape of a fleeing suspect is unreasonable unless the officer has probable cause to believe the suspect poses a significant threat of death or serious physical injury to the officer or others. A Memphis officer had shot 15-year-old Edward Garner in the back of the head as he climbed a fence fleeing a residential burglary, under a statute permitting all necessary means to arrest any fleeing felon. The Court held the statute unconstitutional as applied.

Two features of Garner get misused. The first is the warning requirement, which is conditional: a warning is required "where feasible." Feasibility is a fact question, which makes it an expert question. Was there time, distance, cover, and an audible path for a warning? Body-worn camera audio, radio traffic timestamps, and shot timing are the evidence. "Not feasible" is a conclusion that has to be built from the timeline, not asserted from the witness chair.

The second is that Garner isn't a separate test with its own elements. Scott v. Harris, 550 U.S. 372 (2007), said so directly: Garner "did not establish a magical on/off switch that triggers rigid preconditions whenever an officer's actions constitute deadly force." Garner is Graham applied to one set of facts. An expert who testifies that Garner imposes freestanding elements will get impeached with Scott.

Barnes v. Felix removed the chronological blinders

Barnes v. Felix, 605 U.S. 73 (2025), decided May 15, 2025, made the tactical reconstruction admissible everywhere.

Ashtian Barnes was stopped on a Houston tollway for outstanding toll violations on a rental car. He didn't have his license, he rummaged in the vehicle, the deputy reported smelling marijuana and ordered him out. Barnes restarted the engine and the car began to move. The deputy stepped onto the doorsill of the moving car with his weapon drawn, unable to see inside, and fired twice within about two seconds. The whole stop lasted about two minutes.

The Fifth Circuit had applied its "moment of threat" doctrine, assessing reasonableness only in the two seconds while the deputy was on the doorsill, and excluding as legally irrelevant everything he did to get there. A unanimous Supreme Court rejected that rule.

The inquiry, the Court said, has no time limit. Courts have to consider all the relevant circumstances "including facts and events leading up to the climactic moment," because "earlier facts and circumstances may bear on how a reasonable officer would have understood and responded to later ones."

What that changed, practically: the pre-shooting timeline is now legally relevant in every circuit. Approach, positioning, cover, contact-and-cover discipline, radio and backup decisions, the decision to close distance, and the decision to step into a vehicle's path are all properly part of the totality. Before Barnes, a defense motion in several circuits could strike that evidence by rule. It can't now.

What Barnes didn't decide is the crux, and Part 8 takes it up. The Court wrote that it wasn't addressing "whether or how an officer's own creation of a dangerous situation factors into the reasonableness analysis." Pre-shooting conduct comes in as context. Whether an officer who made the peril himself can lose an otherwise-reasonable shooting is still open.

Fleeing vehicles, and how long the shooting may continue

Most large agencies now prohibit shooting at a moving vehicle except where deadly force is being used against the officer by means other than the vehicle itself. That policy is nearly universal and older than most people assume. In a vehicle case it's usually the strongest single piece of evidence, because the officer's own department told him not to do the thing he did.

Restraint, and why duration is a constitutional fact

Lombardo v. City of St. Louis, 594 U.S. 464 (2021), is the most important prone-restraint decision in the reporter. Nicholas Gilbert, arrested for trespassing and a missed court date, tried to hang himself in a holding cell. Six officers held him face down on the floor, handcuffed behind the back and in leg shackles, with pressure on his back, for about fifteen minutes. His breathing became abnormal and he died.

The Supreme Court vacated and remanded because it couldn't tell whether the court below had treated prone restraint as per se constitutional "no matter the kind, intensity, duration, or surrounding circumstances" so long as the person appeared to resist. It faulted the panel for failing to weigh four things: the pressure on Gilbert's back, that officers had already handcuffed and shackled him, how long they held him, and the department's own training that pressing down on a prone subject can cause suffocation.

That last point is a template for admissible method. The Court itself pointed to the agency's training on the danger. When a department has warned its own officers about a hazard in writing, that document does more work than any outside standard.

Set Lombardo against Rivas-Villegas v. Cortesluna, 595 U.S. 1 (2021). During a domestic call with a reported chainsaw, an officer put a knee on an armed subject's back for about eight seconds. The Court granted immunity. Fifteen minutes on a restrained, suicidal detainee at one pole, eight seconds on an armed subject at the other. The job is to place the incident on that continuum with measurements, not with adjectives.

Which amendment governs, and why it decides the case

Graham says the more specific constitutional provision controls, so the injured person's status picks the amendment. Getting that wrong is the most common analytic error in this field, and it's expensive, because the standards differ enormously in difficulty.

The Kingsley factors, which Lombardo adopted, are worth memorizing because they're the closest thing in this area to a checklist: the relationship between the need for force and the amount used, the extent of the injury, efforts made to temper or limit the force, the severity of the security problem, the threat reasonably perceived, and whether the person was actively resisting.

Each officer, separately

White v. Pauly, 580 U.S. 73 (2017), holds that a reasonable officer arriving late to an ongoing police action may assume that proper procedures, including officer identification, have already been followed. The instruction for the expert is absolute. Analyze each officer separately. In a multi-officer shooting, reconstruct each officer separately: his information set, arrival time, sightlines, radio traffic, and view of the subject. A collective narrative, "the officers knew," is analytically wrong and legally fatal. Reconstruct who knew what, when, and from where.

The eight questions every police reconstruction has to answer

1. Was there a seizure, and of whom? Torres and Lewis mark the boundary. This determines the standard.
2. What was the injured person's status? This determines which amendment.
3. What did each officer individually know, and when? Never a collective narrative.
4. What was the full pre-force sequence? Admissible everywhere after Barnes, framed as bearing on what a reasonable officer would perceive.
5. How much time was actually available? The pivot for Graham's split-second allowance, Garner's warning feasibility, and Lewis's deliberation scale. It's measurable.
6. When did the threat end, and when did firing stop? Plumhoff. Account for perception and reaction lag.
7. What did the video actually capture, and what did it not? Scott. Never opine against plain video. Explain its limits instead.
8. What did the agency's own policy and training say about this exact maneuver? Lombardo. It's the municipal hinge, and it's the most persuasive evidence with juries even where it isn't the constitutional standard.

Sometimes the officer builds the emergency himself. He closes the distance, gives up his cover, wakes a man at gunpoint, shouts, and shoots the man who moves. The threat he answered was one he made.

That's officer-created jeopardy. It goes by officer-created danger, manufactured danger, and provocation too. The names matter, because two of them describe a doctrine the Supreme Court rejected and the others describe an argument that survives. Get the vocabulary wrong in a report and the defense will use it to argue you're advancing a theory the Court killed in 2017.

The doctrinal problem, stated once

Graham asks whether force was objectively reasonable. Courts disagreed for decades about how far back in time that inquiry reaches. Two Supreme Court decisions now define the field, and they cut in opposite directions.

• County of Los Angeles v. Mendez (2017) closed the door on treating a prior constitutional violation as an automatic multiplier that converts reasonable force into excessive force. It expressly opened a different door: ordinary proximate causation.
• Barnes v. Felix (2025) closed the door on temporal blinders. Courts have to look at what preceded the force. It expressly declined to say whether an officer's own creation of the danger counts against him.

These are two distinct routes to recovery with different elements, different defenses, and different immunity profiles. Plead both.

Mendez, and the part everyone forgets

County of Los Angeles v. Mendez, 581 U.S. 420 (2017), came out of a search for a wanted parolee in Lancaster, California. Deputies were told a man and a pregnant woman were living in the backyard. In that backyard stood a wooden shack about seven feet on a side, fronted by a blue blanket hung as a door, with an air conditioner, electricity, and a water hose. Angel Mendez and Jennifer Garcia were asleep on a futon inside. A deputy opened the door without a warrant and without knocking. Mendez, awakened, picked up a pellet gun he kept for shooting rats and moved it so he could sit up. A deputy shouted "Gun." Both deputies fired fifteen rounds in about ten seconds. Mendez lost a leg below the knee. Garcia was shot in the back. The parolee was never on the property.

The district court found the warrantless entry unconstitutional, found the shooting itself reasonable under Graham, and awarded roughly $4 million anyway under the Ninth Circuit's provocation rule. That rule made an otherwise-reasonable use of force unlawful on its own where the officer intentionally or recklessly provoked a violent confrontation by committing a separate Fourth Amendment violation.

A unanimous Supreme Court rejected the provocation rule. It "uses another constitutional violation to manufacture an excessive force claim where one would not otherwise exist." The objective-reasonableness analysis has to be conducted separately for each search or seizure alleged to be unconstitutional. The rule also smuggled the officers' subjective intent back into an objective test.

The question the Court has now reserved twice

City of Tahlequah v. Bond, 595 U.S. 9 (2021), granted immunity to officers who followed an intoxicated man into a garage where he picked up a claw hammer and raised it. The Tenth Circuit had denied immunity on the theory that a jury could find the officers' initial step toward him and their cornering of him recklessly created the situation. The Supreme Court reversed and expressly declined to decide "whether recklessly creating a situation that requires deadly force can itself violate the Fourth Amendment."

Barnes reserved the same question in 2025. Pre-force conduct is admissible everywhere as context bearing on what a reasonable officer would have perceived. Whether recklessness in creating the peril can itself support liability remains undecided at the Supreme Court and contested in the circuits.

That's not a reason to avoid the subject. It's a reason to frame it correctly.

How to say it, and how not to

Where the circuits stand

Barnes abrogated the temporal rule in every circuit that had one. Every "moment of threat" case is now bad law as to the time frame. None of them has been overruled on the substantive question of whether officer recklessness in creating the danger supports liability. Treat pre-2025 circuit law as authority on the substance and as abrogated on the framing.

The seven fact patterns

Officer-created jeopardy isn't a general complaint about tactics. It's a set of recurring, recognizable patterns, each with its own evidence and its own counterweight.

1. Stepping into the path of a moving vehicle, then shooting the driverThe officer eliminates his own escape route and converts a fleeing-vehicle case into a self-defense case. Starks, Abraham v. Raso, Waterman, and Barnes itself. The counterweight is Scott v. Harris and Plumhoff, which are generous to officers on fleeing-vehicle force. The strongest evidence in this pattern is almost always the agency's own policy prohibiting shooting at moving vehicles.
2. Escalating a mental-health or welfare check into an armed confrontationOfficers are summoned to help, and their approach converts a contained crisis into a lethal one. Sevier, Allen, Hastings, Vos, Ceballos. Justice Sotomayor, concurring in Case v. Montana on January 14, 2026, wrote that the presence of law enforcement can escalate a situation rather than calm it. She noted that people with serious mental illness are seven times more likely to be killed in police encounters, and cited 178 cases over two years in which a call for help ended with officers killing the person they came to help.
3. Unannounced or no-knock entry that provokes armed resistance by a lawful occupantThe signature pattern. The occupant's response is lawful self-defense against apparent intruders, and it's then used to justify the force. Mendez is the doctrinal case. Apply the Mendez lesson directly. Immunity usually protects the knock-and-announce violation. The unlawful entry, whether a defective warrant, a wrong address, or none at all, is the predicate that carries proximate-cause damages.
4. Closing distance on an emotionally disturbed person instead of containingDistinguishable from the second pattern by the specific tactical error: closing and cornering rather than isolating, containing, and slowing. Hastings is the paradigm, an officer entering a bedroom and spraying a man holding a sword in a confined space. Allen is the outdoor version.
5. Failure to identify as policeThe occupant's armed response is a lawful reaction to what he reasonably perceives as a home invasion. Claybrook, plainclothes officers in an unmarked car. Young v. Borders, pounding at 1:30 in the morning without announcing. Young v. City of Providence, misidentification of an off-duty officer. Note the trap in White v. Pauly: the shooter may get immunity because he arrived late and could assume proper procedure was followed, and the setup officers may then get immunity derivatively. Plead the setup officers' conduct as an independent entry or seizure violation, not merely as causation feeding the shooter's claim.
6. Positioning failures that eliminate cover and timeDriving directly to the subject. Standing in the open with no barrier. Approaching inside the reactionary gap. Parking so that the only exit is covered. This is the pattern that Barnes made admissible everywhere, and it's the one that's most reliably measurable from video and scene data.
7. Contagious fireOne officer fires and others fire because he fired, not because they independently perceived a threat. The empirical literature supports it. White and Klinger's assessment of multi-shooter incidents in Criminal Justice and Behavior (2012), and an experimental test published in the Journal of Criminal Justice (2024) finding officers were more likely to fire, and fired more rounds, when a peer fired first. The doctrinal hooks are Claybrook's segmenting, Waterman's shots-after-the-threat theory, and the requirement that each officer's force be assessed individually.

Waking, startling, and choosing the stimulus

When officers wake a person, or startle one, they've picked the stimulus. The movement they then shoot at is the movement they chose to provoke.

Disorientation on waking is ordinary human function. A person coming out of sleep can't immediately locate the source of a sound or a light, can't immediately parse speech, and orients by moving and by reaching. That's not unusual behavior requiring explanation. It's the expected behavior, and it's the reason the accepted tactic is to wake a person from a distance, from behind cover, with clear direction and time. Mendez and the shack, and Amir Locke asleep on a couch during a no-knock entry, are the same fact pattern separated by twelve years.

Add a light aimed at the eyes and the problem compounds. A bright light at night blinds a person to everything behind it. He can't tell who is shouting, can't see a uniform, and can't read the situation he's being ordered to answer. Announcement is the standard remedy and it costs one word.

The command carries the same problem. Compliance takes four steps: perceiving the command, understanding it, choosing a response, and moving. Each step takes time. A command that doesn't tell a man holding an object what to do with it isn't a complete instruction. Contradictory commands from several officers are worse than none. Add a language barrier and the interval stretches further. Where an interval between command and shot is roughly a second, the question isn't whether the person complied. It's whether any person could have.

The training doctrine that supplies the standard of care

Police training has a name for the opposite of officer-created jeopardy, a curriculum that teaches it, and a randomized trial behind it.

• Distance plus cover equals time. The Police Executive Research Forum (PERF) states the relationship in that form in its Guiding Principles on Use of Force (2016). Time is what makes de-escalation possible. Everything else in the curriculum follows from it.
• Tactical repositioning. PERF instructs that officers should reposition to isolate and contain a person, and not to draw a line in the sand. That's the direct doctrinal antonym of officer-created jeopardy, stated as a training instruction.
• Proportionality. PERF asks whether there's another, less injurious option available that will safely and effectively achieve the same objective. It frames proportionality explicitly as a standard above the constitutional minimum, not a restatement of it.
• The critical decision-making model. A five-step loop anchored in the sanctity of human life: collect information, assess the situation and the threats and risks, consider police powers and agency policy, identify options and determine the best course of action, then act, review, and reassess.
• Guiding Principle 8. "Shooting at a moving vehicle should be prohibited unless deadly physical force is being used against an officer or another person by means other than the moving vehicle itself." The New York City Police Department adopted such a policy in 1972. Its officers were involved in nearly a thousand shooting incidents a year. The next year that number fell thirty-three percent, and it has declined steadily since, to fewer than a hundred. State the decline as multi-decade, not as a single-year effect of one policy.
• The reactionary gap. The minimum space needed to react to a threat from a person being questioned or detained. Commonly taught benchmarks run to twenty or twenty-five feet for edged weapons, ten to twelve feet plus weapon length for impact weapons, and five to six feet for empty hands. The governing principle is that closer proximity creates greater danger, not greater control. Officer-created jeopardy is what happens when an officer collapses that gap himself.

The evidence that answers the Monday-morning-quarterback objection

The strongest objection to any officer-created jeopardy opinion is that it's hindsight dressed as expertise, and that slowing down gets officers killed. There's a randomized answer.

The Louisville Metro Police Department implemented PERF's Integrating Communications, Assessment, and Tactics (ICAT) de-escalation curriculum in 2019. Engel, Corsaro, Isaza, and McManus evaluated it in a stepped-wedge randomized controlled trial published in Criminology and Public Policy (2022). Use-of-force incidents fell 28.1 percent, injuries to community members 26.3 percent, and injuries to police officers 36.0 percent.

That last number is the one to put in front of a jury. It converts "the officers should have slowed down" from an opinion into a finding, and it answers the officer-safety framing on its own terms. Training that slows officers down demonstrably reduces force and reduces injuries to the officers themselves.

How to plead it after Mendez and Barnes

1. Plead the antecedent Fourth Amendment violation as its own count. Warrantless entry, defective warrant, unlawful stop, unlawful seizure at the threshold. Choose one that isn't protected by qualified immunity. Mendez's trap was that the panel built its causation analysis on the immunity-barred claim.
2. Plead shooting damages as proximately caused by that violation, with foreseeability and direct relation pleaded expressly, using the language from the Mendez remand.
3. Anticipate superseding cause. The Ninth Circuit on remand rejected the argument that the occupant's own act of holding or moving a weapon breaks the chain.
4. Plead the excessive-force count separately, and invoke Barnes to place the whole encounter arc inside the totality: approach, positioning, commands, failure to identify, elapsed time, available alternatives.
5. Where the reckless conduct is part of the seizure itself, plead it inside the force claim under Allen, Ceballos, and Starks. That's the Mendez-compliant formulation: immediately connected to the seizure, not a prior violation that provoked it.
6. Plead failure to train, with the ICAT evidence and the PERF principles as the standard of care. Allen and Young v. City of Providence are the models.
7. Plead state-law negligence wherever the forum allows pre-shooting conduct into the duty analysis. California is the strongest. Hayes v. County of San Diego, 57 Cal. 4th 622 (2013), holds that pre-shooting tactics and decisions fall inside the negligence duty analysis. Tabares v. City of Huntington Beach, 988 F.3d 1119 (9th Cir. 2021), confirms that California reaches further here than the Fourth Amendment does.
8. Segment forward, not backward. Claybrook and Waterman let you isolate later rounds after the threat ended. Mendez doesn't touch that theory.

An officer can lose the reasonableness question and still win the case. That single fact governs police use-of-force litigation, and it's why the claim against the department is often the only route to a full recovery.

The framework

Qualified immunity asks two questions. Did the facts, taken in the light most favorable to the plaintiff, show a constitutional violation? And was the right clearly established at the time? Saucier v. Katz, 533 U.S. 194 (2001), made the sequence mandatory. Pearson v. Callahan, 555 U.S. 223 (2009), unanimously made it discretionary, and courts now routinely skip the first question and grant immunity on the second alone.

The body of clearly established law can't grow if courts never reach the merits. Every case decided on the second prong alone leaves the next plaintiff with the same gap in the precedent.

On what "clearly established" requires, the Court has been consistent and demanding. Ashcroft v. al-Kidd, 563 U.S. 731 (2011): existing precedent must have placed the question "beyond debate," and courts are "repeatedly told" not to define clearly established law at a high level of generality. Mullenix v. Luna, 577 U.S. 7 (2015): the question is whether the violative nature of particular conduct is clearly established in the specific context of the case. Kisela v. Hughes, 584 U.S. 100 (2018): precedent must "squarely govern" the specific facts, and "specificity is especially important in the Fourth Amendment context."

Where the doctrine stands now

Zorn v. Linton, decided March 23, 2026, is the current statement of the specificity requirement and the first significant use-of-force immunity ruling after Barnes. A protester sat with her arms linked during a statehouse sit-in. A Vermont state police sergeant warned her that force would follow if she didn't stand, then put her in a rear wristlock. The Second Circuit denied immunity relying on a 2004 circuit decision. The Supreme Court summarily reversed, holding that the earlier case lacked the requisite specificity because it had said only that a jury could find gratuitous pain, not that the conduct violated the Fourth Amendment. Justice Sotomayor dissented, joined by Justices Kagan and Jackson.

Three consequences follow, and they hit the expert as hard as the brief writer.

• Appellate opinions that merely deny summary judgment don't clearly establish law. A circuit decision saying a jury could find conduct unreasonable isn't a holding that the conduct was unconstitutional. That shrinks the usable precedent pool considerably, and it's the point to have ready when you're asked on cross whether a given case put officers on notice.
• Pain-compliance and low-level force cases are now harder. If you work protest, crowd-control, or passive-resistance matters, Zorn is the leading obstacle.
• The warning was the pivot. The Court distinguished the earlier case partly because a warning had been given here. As in Garner, whether a warning was given, and whether it was audible, comprehensible, and followed by a meaningful opportunity to comply, is a reconstruction question carrying dispositive legal weight.

Barnes widened what a court may look at. Zorn narrowed the way out. The merits inquiry now takes in the whole encounter, and the immunity screen above it got tighter in the same stretch of terms.

The obvious-case exception, and why not to build on it

Hope v. Pelzer, 536 U.S. 730 (2002), refused to require materially similar facts and held that the standard is fair warning. Officials can be on notice even on novel facts, because a general rule already in the case law may apply with obvious clarity to what they did. Taylor v. Riojas, 592 U.S. 7 (2020), revived Hope after nearly two decades, holding that no reasonable officer could have concluded it was permissible to house an inmate in cells covered in human waste for six days.

The exception is real and it's narrow. It reaches conduct that shocks on its face, prolonged and gratuitous cruelty, not contested split-second force decisions. Don't build a shooting case on it.

What the empirical record actually shows

Two careful studies reach different conclusions, and an honest expert should know both, because whichever one you cite, opposing counsel will hand you the other.

• Joanna Schwartz, "How Qualified Immunity Fails," 127 Yale Law Journal 2 (2017). Reviewed 1,183 Section 1983 cases filed in 2011 and 2012 across five federal districts. Defendants raised immunity in about 37.6 percent of the cases where it was available. Courts granted it about 17.9 percent of the time across all stages, and dismissed only 3.2 percent of all cases on that ground. Regional variation was wide.
• Alex Reinert, "Qualified Immunity at Trial," 93 Notre Dame Law Review 2065 (2018). Reviewed 211 cases that reached a jury with immunity raised. Plaintiffs won just over 25 percent, defendants nearly 70 percent. Reinert's conclusion cuts the other way: immunity can be a powerful barrier when it reaches a jury.

The Reuters investigation published in 2020 analyzed 529 federal appellate rulings since 2005 and found that the Fifth Circuit granted immunity in 64 percent of police requests and the Ninth Circuit in 42 percent, with district-level variation of the same magnitude. Forum matters more in this field than in almost any other.

Monell: the claim that survives the officer's immunity

Monell v. Department of Social Services, 436 U.S. 658 (1978), holds that municipalities are persons subject to suit under Section 1983, but only where the unconstitutional action implements an official policy or a governmental custom. Respondeat superior doesn't reach a municipality.

Municipalities have no qualified immunity, under Owen v. City of Independence, and they can't be assessed punitive damages, under City of Newport v. Fact Concerts. That combination is why the municipal claim is often the only path to full recovery when the individual officer wins immunity, and why the municipal defendant fights it hardest.

There are five routes.

1. Express policy that's itself unconstitutional. Rare, and strongest when present, because the writing is the evidence.
2. Widespread custom or practice so persistent and well-settled as to have the force of law. Requires multiple similar prior incidents, and courts demand genuine factual similarity rather than thematic resemblance.
3. A single decision by a final policymaker. Pembaur v. City of Cincinnati, 475 U.S. 469 (1986). Title alone doesn't establish final policymaking authority. You have to identify the source of final authority in that specific subject area under state and local law.
4. Ratification. City of St. Louis v. Praprotnik, 485 U.S. 112 (1988). A final policymaker's approval of both the subordinate's decision and the basis for it. Courts are skeptical, and a failure to discipline after the fact isn't enough on its own. The strong version is an affirmative finding that a shooting was within policy on facts that plainly weren't.
5. Failure to train or supervise. City of Canton v. Harris, 489 U.S. 378 (1989). Part 6 treats this claim in full, in both its constitutional and its state-law forms.

Graham sets a floor, not a ceiling. A shooting can satisfy the Fourth Amendment and still violate a state statute, a state constitution, an agency policy, and a criminal statute. This is the part of the field most practitioners get wrong, and it's where a case that loses federally can win.

What the strict-standard states require that the Constitution doesn't

Since 2018, a dozen or more states have legislated above the constitutional minimum. These statutes share six features. They demand necessity, not just reasonableness. They make officers try de-escalation, exhaust less-lethal alternatives, warn, intervene, and render aid.

State causes of action where qualified immunity isn't available

Four states have entirely barred qualified immunity as a defense in their own courts: Colorado, Montana, Nevada, and New Mexico. Several others have limited it, and one commonly cited group has been misdescribed.

• Colorado. Revised Statutes section 13-21-131 creates a state cause of action against peace officers for state constitutional violations, including failure to intervene, and states that qualified immunity isn't a defense. Two-year limitations period. Mandatory attorney fees for a prevailing plaintiff. The employer indemnifies. But if the employer finds the officer didn't act in good faith and reasonably, that officer owes the lesser of five percent of the judgment or twenty-five thousand dollars.
• New Mexico. The New Mexico Civil Rights Act (2021) creates an action for deprivation of state constitutional rights, bars any public body or person from asserting qualified immunity, waives sovereign immunity, and caps damages at two million dollars per claim. Claims run against the public body, not the individual officer, which changes how you caption the case.
• Nevada and Montana. Both by judicial decision rather than statute. Mack v. Williams (Nev. 2022) and Dorwart v. Caraway (Mont. 2002).
• Connecticut, correctly stated. General Statutes section 52-571k creates a state cause of action but preserves a defense where the officer had an objectively good faith belief the conduct was lawful. Sources that say Connecticut eliminated qualified immunity are imprecise. One-year limitations period.
• Massachusetts, correctly stated. The 2020 act didn't remove the clearly-established requirement. Immunity is stripped only where the officer's conduct results in decertification.
• New York City. A municipal cause of action barring qualified immunity or any substantially equivalent immunity for city officers. City level only. New York State law is unchanged.

Movement runs the other way too, and it's understated. Iowa codified qualified immunity with the federal clearly-established standard in 2021. Louisiana expanded immunity in 2024, eliminating negligence as an exception. Alabama expanded it in 2025.

How to plead it, and what the expert writes

Plead the state statutory violation as a separate count. Don't fold it into the Section 1983 count. Three reasons.

1. Qualified immunity is a defense to the federal claim only. It has no application to a state statutory or common-law claim, and in the abolition states it's barred even in the state constitutional action.
2. A statute requiring de-escalation, exhaustion of alternatives, a warning, or an articulable specific threat imposes duties the Fourth Amendment doesn't. A defendant can win the federal count and lose the state count on the same facts.
3. The statute supplies a negligence per se or statutory standard-of-care theory in states that recognize one, which converts an intentional-tort case carrying a self-defense privilege into a duty case.

The expert's job here is to keep the standards apart. An opinion that force was unnecessary is a state-law opinion in California, Washington, Colorado, Maryland, Minnesota, Virginia, and Oregon. It isn't a Fourth Amendment opinion, because the Fourth Amendment asks about reasonableness, not necessity. An expert who slides between the two invites a Rule 702 motion and gives the defense a clean cross-examination. Structure the report so that one opinion addresses agency policy, one addresses the state statutory standard, one addresses generally accepted police practices, and the ultimate constitutional question is left to the jury. Part 16 sets out the full division between what a firearms expert may say and what belongs to counsel.

In federal court, expect a motion to exclude testimony on state statutes and policy as irrelevant to the constitutional claim. Answer that policy and training evidence bears on what a reasonable officer in this defendant's position would have known and done. That's the Graham question itself.

Most excessive-force litigation doesn't involve a shot. It involves a person held face down until he stopped breathing, a neck restraint, a conducted-energy device applied to someone already handcuffed, or a device drawn in place of the one the officer meant to draw.

Each of these carries its own medical causation question, its own governing standard, and its own body of training and policy. Where a firearm was used, the volley is its own analysis. The number and spacing of the rounds is a separate question from the decision to fire the first round.

Prone restraint and positional asphyxia

A person held face down, with weight applied to the back, and with the arms restrained behind the body, can't expand the chest normally. Compression of the chest and abdomen against the ground, combined with downward force, restricts the movement of the diaphragm. Where the person has been struggling, oxygen demand is already elevated and the reserve is already spent. Add restraint of the arms, leg restraints, and obesity or intoxication, and the margin disappears.

This isn't a contested finding. It's in the training materials of most agencies in the country, and that's what makes it litigable. Lombardo turned on exactly this: the Supreme Court pointed to St. Louis's own training that pressing down on a prone subject's back can cause suffocation. When a department has warned its own officers in writing about a hazard, that document outperforms any outside standard, because it establishes both the standard of care and the notice.

Set the poles. Fifteen minutes of prone restraint on an already-handcuffed and shackled detainee, in Lombardo. Eight seconds of a knee on the back of an armed subject during a violent domestic call, in Rivas-Villegas, where immunity was granted. Neither case decides the one in front of you. The job is to place the incident between them with measurements, and to resist the temptation to characterize instead of measure.

Watch the causation fight. The defense will point to intoxicants, cardiac disease, or an excited-delirium theory. The medical examiner's findings and any revision to them become central, and revisions happen: in one widely reported Washington matter the examiner initially cited methamphetamine and cardiomegaly as contributors and later concluded restraint alone caused death by hypoxia. Where a chemical restraint went in on scene, measure the dose against body weight. The claim can then reach the emergency medical personnel as well as the officers.

Neck restraints

Two mechanisms get confused and they aren't the same. A respiratory or airway hold applies pressure to the front of the neck and restricts breathing. A vascular or carotid hold applies pressure to the sides of the neck and restricts blood flow to the brain. The first is dangerous and slow. The second renders a person unconscious in seconds and can kill.

Policy has moved sharply. A study of use-of-force policies across the hundred largest cities found that departments prohibiting chokeholds rose from about twenty-two percent to about ninety-two percent since 2015 and 2016, and that departments imposing a duty to intervene against excessive force rose from about twenty-nine percent to about ninety-three percent. At the state level, roughly nine states plus the District of Columbia have enacted complete bans and about eight more restrict the technique to circumstances where deadly force would be justified. Massachusetts bans it outright with no deadly-force exception. Colorado defines the prohibited technique by mechanism, covering both the airway and the vascular versions.

So the rule that governs is jurisdiction-specific and date-specific. Pull the policy version in force on the date of the incident, not the current one, and pull the versions on either side of it. Post-incident policy changes tell you what the department knew was wrong.

Weapon confusion

An officer intends to draw a conducted-energy device and draws a pistol instead. It's documented, it's recurring, and it has produced criminal convictions.

The mechanism is a capture error. The officer draws the firearm constantly and the device rarely. Under cognitive load from threat assessment, time pressure, and simultaneous verbalization, the practiced draw captures the unpracticed one. Howard Williams examined twenty such shootings from 2001 forward in "Weapon Confusion: TASER CEWs, Firearms, and Human Error Theories," Criminal Justice Review (2022), applying Norman's theory of human action and Reason's error taxonomy, and found evidence of both capture errors and action-execution slips. He concluded that policy and training changes haven't eliminated the error. Fixing it takes simultaneous change in weapon design, policy, training, and procedure.

The litigation framing matters. Offered as an individual excuse, the capture-error theory is contested and has been criticized as unfalsifiable in the individual case. Offered the other way, it's strong. Capture error is a known, documented, recurring failure mode. So the proximate cause is holster placement, cross-draw policy, weight and grip differentiation, and draw-frequency training. That turns a weapon-confusion suit into a training and policy claim against the department rather than a claim against one officer. In the most prominent modern example, the officer who made the error was a twenty-six-year veteran and a field training officer. No fact in this line of cases does more work.

One caution on the physical-differentiation argument. Color, weight, grip, and holster position are the standard answers, and the literature suggests they're insufficient under stress, because attentional narrowing suppresses awareness of exactly those cues. Don't overclaim in either direction.

Mistaken-object shootings

A replica, an airsoft or pellet gun, a cell phone, a wallet, a garden tool. The recurring question is whether the officer identified the object reasonably, and the answer is usually yes.

A licensed replica of a service pistol is black, carries a removable magazine, and has a moving slide. At night, through automotive glass, at a glance, nobody distinguishes it from the pistol it copies. Say so when it's true. A pellet or airsoft gun missing its orange tip is functionally indistinguishable at distance in poor light. The hindsight bar in Graham means the object's true nature isn't part of the reasonableness calculus at all.

That concession costs nothing and it's worth a great deal, because it isolates the real question. The appearance of the object usually isn't what makes a shooting unreasonable. What makes it unreasonable, where it's unreasonable, is the sequence of choices that made a snap identification necessary in the first place: the distance closed, the cover given up, the stimulus chosen, the time not taken. Every one of those choices was made before the object was ever seen.

The object's true nature matters elsewhere in the case. It matters to the municipal claim, because training on replica and air-powered guns, and on distinguishing them from firearms, is a documented curriculum subject. It matters to damages. And it matters to the risk that actually existed. That's a different question from whether the officer's perception was reasonable. Label it that way in your report so nobody can accuse you of conflating the two.

The volley: each round is a separate use of force

Analysts skip this one most often, and it's frequently where the case lives.

Plumhoff forecloses a bare complaint that too many rounds were fired: officers justified in firing to end a severe threat need not stop shooting until the threat has ended. Read that sentence carefully, because it contains the whole test. The question isn't the count. The question is when the threat ended, and whether rounds left the barrel after that point.

Eight rounds fired at the rate a service pistol is ordinarily shot spans somewhere between one and two seconds. That interval is measurable to a few thousandths of a second from a body-worn camera audio waveform. Match each round against the wound paths, the trajectory data, and the officers' positions at that instant. The comparison shows where a volley crossed from response into something else.

Reaction lag has to be accounted for honestly and it cuts for the defense. If a person's mean reaction time to a simple, expected stimulus runs around four tenths of a second, then some trailing rounds after a threat ends are physiologically explainable. That research sets a boundary. It doesn't excuse everything past the boundary, and the distance past it is a measurement.

Two doctrinal hooks carry this theory. Waterman v. Batton granted immunity for the initial rounds and denied it for rounds fired after the vehicle had passed the officers, because once the immediate threat was eliminated any continuing belief in serious harm would be unreasonable. Claybrook v. Birchwell holds that multiple discrete uses of force require a separate reasonableness finding for each temporal segment. Neither theory is touched by Mendez, and both survive Barnes.

Contact and cover, and the second officer

In a two-officer contact, the roles are distinct and taught as such. The contact officer handles the interaction. The cover officer maintains position, watches hands and the surroundings, keeps a clear field, and calls a hold when the contact officer's position deteriorates.

So the cover officer carries duties the contact officer doesn't, and his view, his position, and his chance to act are separate factual questions. What did he see. What did he say. What could he have done. Under White v. Pauly the analysis runs officer by officer anyway. The duty to intervene gives that inquiry an independent legal home. A dozen or more states now codify it, and more than ninety percent of large-city departments carry it in policy.

Note the federal template. The most-cited modern duty-to-intervene convictions came out of a restraint death. Three of the officers never touched the man. The federal government convicted them for failing to stop the ones who did.

Conducted-energy devices and chemical agents

Both are classified as less-lethal, and both kill people under identifiable conditions. The same issues recur. An officer deploys against a person already restrained or already compliant. He runs repeated cycles. He hits a prohibited target area. He fires on a person in an elevated position or near flammable vapor. And he uses the device to inflict pain rather than to overcome active resistance.

Policy has tightened. Illinois restricts deployment and prohibits targeting the head, chest, neck, groin, and anterior pelvis. Several states require an order and an opportunity to comply before a chemical agent is deployed against an individual, and distinguish individual deployment from crowd-control use. Zorn is the current obstacle for pain-compliance claims at the federal level, and the state statutes are where these cases will increasingly be litigated.

Sometimes the person shot is not the person the shooter aimed at. A neighbor behind a wall. A store clerk at the register. A hostage. A child asleep in the next room. A man walking to dinner.

These cases look like the rest of this guide and litigate nothing like it. The bystander has no Fourth Amendment claim in most of the country, because he wasn't seized. What's left is a harder constitutional theory, a state tort claim, or, against a private guard, the strongest posture any of these plaintiffs ever gets.

One clarification before the doctrine. This Part is about the bystander who was hit. It isn't about the bystander who watched. Negligent infliction of emotional distress, the claim of the parent who saw the shooting and wasn't touched by it, runs on the zone of danger rule or the Dillon v. Legg factors, and Part 5 covers it. The two claims share a word and share nothing else.

The seizure problem decides everything else

Brower v. County of Inyo, 489 U.S. 593 (1989), holds that a Fourth Amendment seizure happens only through means intentionally applied. Read the sentence, because the Court picked the bystander as its example of a person who isn't seized.

Brower adds the illustration that every defense brief quotes: if a parked police car slips its brake and pins a passerby against a wall, "it is likely that a tort has occurred, but not a violation of the Fourth Amendment."

Torres v. Madrid, 592 U.S. 306 (2021), requires force applied "with intent to restrain" and says accidental force won't qualify. It doesn't say whether that intent has to run to the person struck. There's no footnote on bystanders, no dictum, nothing. Roxanne Torres was the obvious and only target. Both sides now argue from the same two sentences and reach opposite results depending on whether the intent must reach the plaintiff or merely somebody.

Where the circuits stand, and the 2026 case that split them

Geometry sorts these cases better than circuit maps do

Line the outcomes up by where the person was standing, and the pattern is sharper than the circuit split suggests.

The bystander the officer saw and fired toward anyway is the strongest configuration, and no case yet holds that seeing him converts the strike into a seizure of him. It still does four kinds of work. It's the whole fight under Kilnapp's objective test, because an officer who fires into a space he knows is occupied has manifested an intent to apply force there. It turns an accident into knowledge-based intent under ordinary tort principles. It attacks Claybrook's protective language, which is expressly conditioned on the circumstances being unforeseen. And it decides the state negligence claim outright, because foreseeability of the bystander is the whole duty question.

The unseen bystander behind a wall or a door is the hardest configuration of all. Claybrook, Rucker, and the rounds that crossed a shared wall in Louisville all involved invisible victims, and all of them failed. That plaintiff's route isn't the Fourth Amendment. It's the target and backdrop standard of care set out below, carried by state tort law.

The Fourteenth Amendment route, and why it almost never works

County of Sacramento v. Lewis leaves the true bystander one federal theory. Negligence is never enough. Deliberate indifference may shock the conscience where actual deliberation was practical. Where officers had to make split-second judgments, only a purpose to cause harm unrelated to a legitimate law enforcement objective will do.

Every circuit to reach a bystander shooting has applied the purpose-to-harm tier, on the theory that an armed confrontation is the paradigm of a situation demanding instant judgment. Claybrook supplies the sentence the defense will quote: "even precipitate recklessness, fails to inch close enough to harmful purpose."

The tier isn't fixed by the word "shooting." It's fixed by whether deliberation was practical. A six-hour barricade, a planned warrant service, or a static perimeter isn't a stairwell ambush. That's the open ground, and no appellate court has yet closed it.

State-created danger is worth pleading in one configuration and no other. The elements vary by circuit. What kills the theory is the requirement that the plaintiff be a foreseeable victim rather than a member of the public at large. It has traction only where officers knew of a specific person or a discrete group and fired anyway: the neighbor whose apartment they knew was occupied, the crowd they could see behind the target, the child they knew was in the room. It supplements the Lewis claim. It doesn't replace it.

Where bystanders actually recover

Not in federal court, on the whole. The survey found no federal appellate decision awarding a bystander damages under substantive due process for an inadvertent police shooting.

They recover under state law, and the numbers are real.

• Denver, October 17, 2025. $19.75 million. An officer fired at an armed man at bar close in Lower Downtown and struck six bystanders. The jury returned the verdict against the former officer individually. The City and County of Denver was never named. He had already pleaded guilty in 2024 to third-degree assault, drawn eighteen months of probation, and lost his peace officer certification. Colorado is the jurisdiction where this result is most available, because its state cause of action makes qualified immunity unavailable and mandates fees to a prevailing plaintiff.
• Los Angeles, August 30, 2024. $9.5 million. An officer's round killed a Trader Joe's assistant manager standing inside the entrance during an exchange of fire with a fleeing suspect. The district attorney had found the officers justified in 2020. Counsel called it the largest pretrial settlement the city had paid in a police shooting.
• Riverside, October 27, 2021. $17 million. An off-duty officer shot a man in a Costco and wounded both of the man's parents, who were unarmed bystanders. The federal jury found the officer acted within the scope of employment, which put the verdict on the city. The criminal case ended in a mistrial in January 2024 and wasn't retried.

Read those three together. In each one the officer was cleared criminally or never charged, and the money came anyway, because the standard that governs a bystander's tort claim isn't the standard that governs the decision to fire.

Federal agents, including the Secret Service

Start by writing off the constitutional claim. Egbert v. Boule and Goldey v. Fields have cut Bivens to almost nothing outside its three original contexts, and a bystander shot by a federal agent presents a new context on every axis: the agency, the injury mechanism, and the plaintiff's status as a non-suspect. Plead the Federal Tort Claims Act and preserve Bivens for appeal.

The Federal Tort Claims Act is the real route, and its mechanics decide these cases.

1. State law supplies the standard. Title 28 of the United States Code, section 1346(b)(1), makes the United States liable where a private person would be liable under the law of the place where the act occurred. The comparison runs to a private individual, not to a municipal officer, so state police privileges don't transfer cleanly.
2. No punitive damages and no jury. Section 2674 bars punitives. Section 2402 tries the case to the court. In a bystander injury case, where the facts carry the emotional weight, losing the jury is a serious structural loss.
3. The discretionary function exception is where the case lives or dies. Section 2680(a) bars claims based on a discretionary function, whether or not the discretion is abused. The plaintiff's move is to find a mandatory, non-discretionary directive the agent violated, which removes the element of judgment. Agency use-of-force policy with prohibitory language is the target. Policy phrased in "should" or "consider" preserves discretion and the exception bars the claim.
4. The law enforcement proviso restores the intentional torts. Section 2680(h) bars claims arising out of assault and battery, then restores six of them, battery included, for acts of federal investigative or law enforcement officers. It doesn't create a cause of action. It removes a bar.
5. Martin v. United States, 605 U.S. 395 (2025), decided the interaction. A unanimous Court held that the Supremacy Clause is no defense to a Federal Tort Claims Act suit, and that the law enforcement proviso overrides only the intentional-tort exception in its own subsection. It doesn't override the discretionary function exception. So plead around section 2680(a) first. The proviso never matters until you do.
6. Exhaust, and watch the clock. Present the claim to the agency in writing with a sum certain before filing. Standard Form 95 is the usual vehicle. Two years from accrual to present, six months from a written denial to sue. The sum certain is the classic malpractice trap.
7. Sequence the claims carefully. Under the judgment bar at section 2676, and Brownback v. King, 592 U.S. 209 (2021), losing the tort claim can extinguish a companion constitutional claim. The Court reserved whether the bar reaches claims in the same lawsuit.

The Secret Service carries firearms and makes arrests under 18 United States Code section 3056, and that statute sets no use-of-force standard. Policy does. The governing document is the Department of Homeland Security's use-of-force policy statement, revised February 6, 2023, and it contains a sentence written for exactly this chapter: before using deadly force at the operator of a moving conveyance, the officer "must take into consideration the hazards that may be posed to law enforcement and innocent bystanders by an out-of-control conveyance." That mandatory phrasing is the kind of language a plaintiff needs to defeat the discretionary function exception. No Secret Service directive of its own is public. Getting it takes a records request or discovery.

The best federal result located is a bench trial. In a failed hostage rescue in Houston, an agent fired a rifle through a double-paned window and killed the bound kidnapping victim. On March 3, 2025, the court awarded $1.911 million plus $475,000 in fees, finding the agent negligent and finding the agents' account that the victim grabbed the weapon to be, in the court's words, intentional falsehoods contradicted by the physical evidence.

One current matter shows how these begin. In May 2026 a man opened fire at a White House checkpoint and Secret Service personnel returned fire and killed him. A soldier walking to dinner was struck in the abdomen and required multiple surgeries. Public reporting stated plainly that it wasn't clear whose round hit him. That's the ordinary state of the record at the outset of a bystander injury case, and it's why attribution work isn't decoration. Attribution is the case.

Private security: the best posture a bystander gets

A guard on private premises exercising a shopkeeper's privilege or ordinary self-defense isn't a state actor, so Section 1983 doesn't reach him. Commissioning as a special police officer, deputization, or joint action with responding police can flip that, and off-duty sworn officers working security are their own contested category.

Losing Section 1983 is the plaintiff's gain here, and it isn't close. Against a private guard the bystander gets:

• Negligence instead of a seizure requirement. Brower, Torres, Rucker, and Claybrook are irrelevant. The entire problem at the top of this Part disappears.
• No qualified immunity. No clearly-established inquiry, and no interlocutory appeal to stall the case for two years. A $32.9 million verdict can't be taken away the way the Baltimore County verdict was.
• No municipal policy requirement. Respondeat superior reaches the employer directly.
• No tort claims act. No short jurisdictional notice deadline, no damages cap, no discretionary-act immunity.
• Punitive damages, a jury, and insurance. Licensed armed security firms carry liability coverage, and many states condition licensure on proof of it.

The price is proving negligence rather than invoking a constitutional standard. In a bystander injury case that's a bargain, because negligence is the lower bar. The officer defendant is judged by objective reasonableness or by whether his conduct shocks the conscience. The guard is judged by ordinary care.

The theories are the familiar ones. Respondeat superior, which the defense contests mainly to break the chain to the employer and defeat punitive exposure. Negligent hiring, training, retention, and supervision, which is direct corporate liability independent of scope. Premises liability, where in many states the duty to select a competent security contractor is personal and non-delegable, so hiring a contractor doesn't insulate the property owner. And in some jurisdictions a third-party beneficiary claim under the security contract.

Post orders are the single most productive discovery target in a guard case. Deviation from written post orders is the cleanest breach evidence available.

Does a justified shot excuse hitting the bystander?

No, and this is the most important rule in the Part.

Restatement (Third) of Torts: Intentional Torts to Persons section 26(c) states the modern rule. An actor privileged in self-defense or defense of a third person who unintentionally harms a bystander isn't subject to intentional tort liability to that bystander. In that circumstance his liability "is determined solely according to applicable principles of negligence."

So the privilege transfers to the intent element and not to the care element. It kills the battery claim. It leaves the negligence claim standing.

The defense will try to collapse the two questions into one: he was entitled to shoot, so he owes nobody anything. They're separate questions. Whether he was entitled to shoot at the assailant turns on the threat the assailant posed. Whether he was negligent as to the bystander turns on what he knew or should have known about the backdrop, the crowd, and the wall behind the target. It also turns on whether a reasonably careful armed person would have fired at all, from that position, at that angle, with that ammunition.

Criminal law splits three ways on the same question, and the forum state's answer decides whether there's any criminal exposure at all.

1. Justification transfers completely. Commonwealth v. Fowlin, 551 Pa. 414, 710 A.2d 1130 (1998). A man blinded by pepper spray and facing a drawn handgun fired eleven rounds and hit a bystander. Held: justified self-defense negates recklessness itself, so the charges requiring recklessness fail. The court limited it: liability returns if the belief was unreasonable, if he provoked the confrontation, or if he could have retreated safely.
2. Justification defeats murder, recklessness liability survives. Commonwealth v. Santana-Rodriguez, SJC-13753 (Mass. Oct. 14, 2025), the modern trend. The justification carries over and the intent follows the bullet, so murder can't rest on the accidental killing of a bystander. But involuntary manslaughter survives where the defensive conduct was wanton or reckless enough to create a high degree of likelihood of substantial harm to an unintended victim. The court expressly rejected the blanket approach.
3. Justification doesn't transfer at all. Model Penal Code section 3.09(3): where the actor is justified but "recklessly or negligently injures or creates a risk of injury to innocent persons, the justification afforded by those Sections is unavailable in a prosecution for such recklessness or negligence towards innocent persons." Note the reach of "creates a risk of injury." Under the Code the endangerment charge is available even where nobody was hit. Most code states have an analogue. Check the forum's justification statute.

The physics that makes the bystander foreseeable

Rule 4 has two halves and the second one is what these cases are about. Be sure of your target, and what's beyond it. Identifying the target satisfies half the rule. The shooter also has to account for what's behind and around it, because the round won't stop at the target if he misses, and often won't stop at the target even if he hits.

Five established facts turn that rule into a foreseeability and breach opinion. None of them is controversial, and every one is taught in the first hour of any firearms course.

Misses are the norm, not the exception.

The RAND Corporation evaluated New York City Police Department firearms training in MG-717 (2008), covering shootings from 1998 through 2006. The average hit rate was 18 percent in gunfights where the subject returned fire and 30 percent where fire wasn't returned. Donner and Popovich, studying Dallas Police Department shootings from 2003 through 2017, found a 35 percent bullet-level hit rate, with officers striking the subject at least once in only 54 percent of incidents. Departments have rarely achieved a 50 percent hit rate since the 1970s.

Put plainly: roughly two of every three rounds an officer fires go somewhere other than the intended target, and in a two-way gunfight it's closer to four in five. Every one of those rounds goes somewhere. That isn't an aberration to be excused. It's the published, department-documented baseline against which a decision to fire in a populated place has to be judged.

Hits don't stop rounds, and walls don't either.

The Federal Bureau of Investigation's wound ballistics work sets the ammunition standard, and it's worth stating exactly what that standard requires. The Bureau's position, stated in 1989 and still governing, is that penetration up to 18 inches is preferable and a handgun bullet must reliably penetrate a minimum of 12 inches of soft tissue. An adult male chest is roughly 9 to 11 inches front to back.

Duty ammunition is deliberately selected to penetrate more than the depth of a human torso. Overpenetration isn't a malfunction. It's a designed-in property of every service round in American law enforcement, and the agency knew it when it chose the load.

The Bureau's test protocol makes the point again. One of the six required events fires through two pieces of half-inch gypsum board set three and a half inches apart, with the gelatin block 18 inches behind, which is a deliberate simulation of an interior residential wall. Others fire through steel sheet, three-quarter-inch plywood, and laminated automotive glass at a compound angle. In every one of them the round has to reach the gelatin and perform. Ammunition that failed those barriers wouldn't be adopted.

Published wall-penetration testing sorts the ammunition types cleanly. Full metal jacket is the worst offender in every caliber. In one test series, 9mm and 5.56 full metal jacket each cleared a simulated interior wall and an exterior wall in the same shot. So did .45 full metal jacket, 00 buckshot, and a 12 gauge slug. Jacketed hollow point substantially reduces wall penetration, because expansion on the first barrier bleeds energy and widens the frontal area. Counterintuitively, a .223 in an expanding or fragmenting defensive load penetrates residential construction less than 9mm full metal jacket, because the light, fast bullet destabilizes and comes apart. Expect the defense to make that point. It's correct as to those loads and wrong as to 55 grain full metal jacket, which is common patrol rifle ammunition.

Absent case-specific testing, assume duty ammunition defeats a hollow-core interior door, a gypsum partition, automotive sheet metal, and automotive glass, and keeps lethal energy past each. The exterior wall is the real variable, and brick, block, or stucco changes the analysis and requires testing.

Hard surfaces redirect rounds into the space where people stand.

A ricocheting bullet doesn't bounce away at the mirror angle the way light does. It leaves at a shallower angle than it arrived, so it travels close to and roughly parallel with the surface it struck. On pavement or a floor that keeps the round in the horizontal plane human beings occupy, sometimes for a long way. A round fired into the ground short of a target doesn't bury itself. It skips downrange at knee-to-chest height.

Critical angles are low. Published figures put drywall near 10 degrees, cement near 11, rough concrete near 11, thin sheet metal as low as 3, and float glass in the teens for common handgun calibers. Below those angles the round deflects rather than penetrating. A grazing ricochet loses relatively little velocity and stays lethal, while a steeper impact can crater and throw the round back out at an angle greater than it arrived, which is unpredictable. Handguns ricochet more readily than rifles, so the standard duty sidearm is the more ricochet-prone weapon on the belt.

Ricochet leaves a signature on the body. Hlavaty and colleagues examined more than 150 entrance wounds from rounds deflected off concrete, asphalt, aluminum sign, brick, and drywall. Every caliber and every surface produced atypical wounds: irregular in size or shape, missing the marginal abrasion, or carrying other injuries. The surface the round struck was the largest factor in how the wound looked. An irregular entrance without a clean abrasion collar is affirmative evidence of ricochet, and it points back to the surface, the trajectory, and the shooter's position.

The danger space runs past a mile.

A 9mm round can travel roughly 2,000 yards, a .45 about 1,500, and a .223 or 5.56 in the range of 3,000 to 3,500. Terminal energy makes the point sharper than distance does: a .223 retains more energy at 500 yards, roughly 330 to 385 foot-pounds, than a 9mm carries at the muzzle. A patrol rifle round that misses stays lethal across a third of a mile of city.

Contagious fire multiplies rounds without improving aim.

DeCarlo, Dlugolenski, and Myers put 169 officers between two confederates and ran a randomized test published in the Journal of Criminal Justice in 2024. When the confederates held fire, 49 percent of participants fired, averaging 1.13 rounds. When the confederates fired, 92 percent fired, averaging 3.21 rounds. Officers were roughly eleven times more likely to fire and fired about 72 percent more rounds.

The stimulus in that experiment was a man pulling a cellphone from his waistband. So the officers who fired were, by construction, firing at a non-threat. Contagious fire multiplies rounds precisely where the target identification was wrong, in an environment the second and third shooters never independently assessed. More rounds, no better accuracy, no independent assessment. That's the recipe for a bystander strike, and it's why the Empire State Building volley in August 2012 wounded nine people with sixteen rounds, all of them struck by police rounds or fragments, as the department itself reported.

Attribution is the case

Before any of the doctrine matters, somebody has to prove whose round it was. The defense position in almost every one of these matters is that the assailant fired the round, or that it can't be determined. Six lines of proof answer it.

Working the bystander case

The two tables below come from matters reported in the public record, with notes after each. Several were unresolved as this edition went out. Verify the current procedural status of any of them before relying on it in a filing.

Private actors: security and civilian

The loss-prevention shooting.

The San Francisco matter is the cleanest illustration of the reasonableness question stripped to its elements. Surveillance video showed a struggle, a release, and the suspect stepping back toward the guard before the shot. No weapon was recovered. The prosecutor's declination turned on whether the guard's stated fear was reasonable under the totality of the circumstances, the same standard that governs the civil claim. The human-factors work here is a frame-by-frame reconstruction of distance, movement direction, and elapsed time between release and shot, matched against the guard's training in the same decision points.

The parking-lot escalation.

The Portland matter shows how quickly a use-of-force case pulls the employer into the frame. A jury didn't stop at judging the guard's split-second decision. It found the property manager, the owner, and the security company independently negligent, reportedly on a theory that the security program's culture and training invited a confrontational response rather than de-escalation. This is the pattern to watch for at intake: when the volume of force is disproportionate on its face, the case usually isn't only about the shooter.

The physical takedown that becomes a shooting.

The Albuquerque matter began as a detention, not a shooting: a tackle, a warning, chemical spray, and only then a firearm, during a struggle over merchandise worth less than the cost of a modest dinner. That sequence matters twice. It shows an escalating decision tree in which each step should have been evaluated against the threat actually presented, not against the initial decision to detain. And once the gun came out during a physical struggle, the case picks up a handling-violation thread: was the discharge a deliberate, aimed shot, or one that occurred because a hand or body in contact with the weapon caused an unintentional trigger press.

The employer's exposure.

Cervantez is decades old and still does the foundational work. A retailer's liability for the conduct of its security personnel doesn't end with the guard. Negligent hiring, negligent training, and negligent supervision are independent theories that survive even where the individual guard's conduct might otherwise be excused or justified. In every matter involving a contracted or in-house guard, work this layer as well as the shooting.

Police matters

The table below is organized by the analytical category each matter teaches, not by notoriety. Criminal and civil outcomes are given separately because they diverge constantly, and that divergence is itself one of the lessons.

Why the criminal and civil columns diverge.

They apply different standards to different questions with different burdens. A criminal acquittal means the state failed to prove a crime beyond a reasonable doubt. It says nothing about whether the force was objectively unreasonable under the Fourth Amendment, and nothing about whether the department's training was deliberately indifferent. The Mesa matter produced an acquittal and $9.5 million in civil settlements. In a Michigan matter, the criminal jury hung and prosecutors declined to retry in 2025. In 2026 the court denied qualified immunity on the same record. Don't let either column be read as the answer to the other.

The pattern that recurs across the police table.

Look at what's in the "what it turns on" column. Positioning. Announcement. The stimulus chosen. The interval allowed. The information not relayed. In most of these matters the shooting decision itself, taken in isolation at the moment it was made, is defensible or close to it. What isn't defensible is the sequence that produced the moment. That's the pattern Barnes made admissible everywhere, and it's why the pre-force reconstruction is now the center of the work rather than a preliminary to it.

The information-relay failures.

Two of the matters above turn on what a caller said and what an officer was told. A caller says the gun is probably fake and the dispatcher doesn't relay it. A caller's account of her own role turns out to be false. Neither fact controls the reasonableness of what the officer perceived, because the officer can only act on what he received. Both facts control something else. Did the agency's dispatch protocol and call-intake training meet the standard? That's a municipal question. Should the gap between the reported scene and the found scene have slowed the response? That's a tactical one. Keep the two separate in a report.

The theory pleaded should track the case family and the identity of the shooter. Consult counsel licensed in the relevant jurisdiction for the elements and defenses that apply there.

Against any shooter

1. Assault and batteryThe intentional tort claim against the shooter directly. Available in both case families. In a handling case it's typically pleaded alongside negligence in the alternative.
2. Negligence, against the shooterThe claim of choice in an unintentional-discharge case: the shooter owed a duty of reasonable care in handling a firearm and breached it. The four cardinal rules function as the standard-of-care benchmark. In several states, notably California, the negligence duty reaches pre-shooting tactical conduct that the Fourth Amendment analysis reaches less readily.
3. Wrongful death and survival actionsTwo different claims with two different owners, and conflating them costs money. The survival action belongs to the estate and carries the decedent's own damages: conscious pain and suffering before death, medical expenses, and in many states punitive damages. The wrongful death action belongs to the statutory beneficiaries and carries their losses: financial support, services, and loss of society and companionship. Which damages are available under each, and which are capped, is entirely a matter of state statute. Plead both.
4. Negligent infliction and bystander claimsA person who witnessed the killing may have a claim in their own right. State rules on the required relationship and proximity vary widely.

Against a private employer or premises owner

1. False imprisonmentAttaches to the detention that preceded the shooting, independent of whether the shooting itself was justified. A detention that exceeded the shopkeeper's privilege in scope, duration, or force can support liability even where the eventual use of the firearm is separately litigated.
2. Negligent hiring, training, retention, and supervisionThe employer-layer theory established in cases like Cervantez. Discovery targets the guard's licensing status, firearms qualification records, use-of-force policy, and any prior incidents the employer knew or should have known about.
3. Respondeat superior and vicarious liabilityPuts the security company and, depending on the contract structure, the premises owner on the hook for the guard's conduct within the scope of employment. It reaches a private employer and not a municipality. That gap often makes the security company the better target.
4. Premises liabilityRuns against the property owner independently of the employment relationship with the guard, on the theory that an inadequately vetted or armed security presence made the premises unreasonably dangerous.

Against a police officer or agency

1. Fourth Amendment excessive force, under Section 1983The federal civil rights claim against a state actor, alleging an unreasonable seizure under the Graham standard. It doesn't reach a private security guard or an armed civilian absent state action, and it's the defining feature that separates a police matter from the rest of this guide. Qualified immunity is the answering defense.
2. Unlawful entry, unlawful stop, or unlawful seizure, pleaded separatelyThe Mendez predicate. Plead it as its own count, and pair it with a proximate-cause allegation that carries the shooting damages. That's the surviving route for officer-created jeopardy facts. Choose a predicate that isn't itself protected by qualified immunity.
3. Failure to interveneAn independent claim against officers who observed excessive force and didn't stop it, now codified in a dozen or more states and adopted by policy in the large majority of major agencies. It's also the theory behind the most-cited modern federal convictions of officers who didn't apply the fatal force themselves.
4. Failure to render medical aidDeliberate indifference to a serious medical need, and in several states a statutory duty. The interval from injury to first aid is measurable from video and is often long.
5. Municipal liability under MonellExtends liability to the municipality where a policy, custom, ratification, or a failure to train caused the constitutional violation. No qualified immunity applies and no punitive damages are available. Part 9 sets out the five routes and the proof.
6. Fourteenth Amendment loss of familial associationThe family's own claim, distinct from the estate's Fourth Amendment survival claim, for deprivation of the familial relationship. It applies the shocks-the-conscience framework. Which standard governs, deliberate indifference or a purpose to harm, depends on whether the officers had time to deliberate. Two different standards apply to the same facts, so keep the analysis separate.
7. State constitutional claims where the state provides themIn Colorado, New Mexico, Montana, Nevada, and New York City, a state cause of action exists in which qualified immunity isn't available. After Zorn these tracks are meaningfully more valuable than the parallel federal claim.
8. State statutory use-of-force violationsPleaded as a separate count, never folded into the Section 1983 claim. In the strict-standard states this count can succeed on facts that lose federally.
9. Americans with Disabilities Act and Rehabilitation ActA reasonable-accommodation-during-arrest theory in mental-health cases. It survives Mendez untouched and permits direct scrutiny of tactics.
10. Substantive due process, for a person who wasn't seizedThe bystander's federal theory in most circuits, because a person who wasn't the target of the force usually wasn't seized and has no Fourth Amendment claim at all. It runs on the shocks-the-conscience standard, and whether deliberate indifference or a purpose to harm applies turns on whether officers had time to deliberate. Part 12 sets out the seizure problem, the circuit split, and where bystanders actually recover.
11. Federal Tort Claims Act, against a federal agentThe only live route against federal personnel after Bivens was cut back. State law supplies the standard, there's no jury and no punitive damages, the discretionary function exception has to be pleaded around first, and the administrative claim has to be presented with a sum certain within two years. Part 12 covers the mechanics.
12. State tort claims against the officer and the agencyBattery, negligence, and wrongful death under state law, subject to notice-of-claim deadlines, damage caps, and official immunity. Miss the notice deadline and the state claim is gone regardless of merit. Part 5 sets out these theories in full, along with negligent entrustment, the family's own claims, the culpability ladder, and punitive damages.

The defenses that answer them

Every figure below carries its source and its year. Where a number rests on a single source or a contested methodology, that's stated. Don't cite any of these without checking the current version, and don't cite a figure whose denominator you can't state.

Private security and negligent security litigation

Reliable, current figures on private-security shootings specifically are scarce. Most published data covers negligent security broadly, which includes premises liability for crimes committed by third parties as well as claims against guards themselves.

Two cautions. The older study predates the current wave of retail-theft-driven security shootings and should be treated as a historical baseline, not a current prediction. And the available data understates plaintiff win rates here, because the strong cases settle before trial and never reach a jury-verdict study. The gap between a documented outcome and a case's true settlement value is where many of these matters are won or lost.

How many people police kill, and why nobody agrees

Three counts circulate and they measure different things. Mapping Police Violence counts all deaths from any application of force by any officer, on or off duty, justified or not. Gun Violence Archive codes gunfire-centered incidents and reported 1,404 subjects killed in 2025. The Washington Post's Fatal Force database counted fatal police shootings only, on duty, and stopped being updated in January 2025, covering 2015 through 2024. Any citation to Post data in current work has to say so. Its narrower scope also makes its counts structurally lower and not directly comparable: a restraint death, an off-duty shooting, or an in-custody death falls outside it entirely.

Federal counts are lower still, and the reason is structural. The Post's own methodology statement says it documented more than twice as many fatal shootings a year as federal officials recorded. In 2021 the Federal Bureau of Investigation (FBI) tracked only about a third of departments' fatal shootings. The FBI's National Use-of-Force Data Collection is voluntary and releases data only at participation thresholds. Coroners and medical examiners complete the death certificates, and many of them work for the same county law enforcement structure or sit close to it. They underuse the legal-intervention codes. No comprehensive mandatory federal database exists.

Circumstances, 2025

• Allegedly armed, 76 percent. Unarmed, 22 percent. About 52 percent were reported armed with a firearm and about 2 percent with a vehicle.
• Force type: gunfire 87 percent, vehicle 8 percent, conducted-energy device 8 percent. The categories overlap.
• Mental or behavioral health crisis: 20 percent of people killed showed symptoms, and 89 percent of those deaths followed a response to a call for service.
• 66 percent of all killings followed an officer's response to a call for service rather than a self-initiated stop.

That last pair of numbers is the empirical spine of the officer-created jeopardy argument. Two thirds of these deaths begin with somebody asking for help.

Racial disparity, and the honest caveat

Rates per 100,000 population for 2025, per Mapping Police Violence: Native Hawaiian and Pacific Islander 1.59, American Indian and Alaska Native 0.86, Black 0.75, Hispanic 0.37, white 0.29, Asian 0.08. On that framing Black Americans are killed at roughly 2.6 times the white rate. The Lancet study covering 1980 to 2019 put the figure at about 3.5 times.

State the caveat before opposing counsel does. Per-capita comparison to the general population is the standard framing and it's contested. The competing denominators are population, encounter rate, and crime rate, and the estimated magnitude shifts materially depending on which one is chosen. The disparity is robust across denominators. Its size isn't.

What it costs, and who pays

Current city-level figures show the trend continuing. Chicago paid about $259 million in settlements and verdicts in 2025, more than triple the 2024 figure of $107.5 million, plus a record $36 million to outside counsel. New York City paid about $117 million to resolve police misconduct suits in calendar 2025. That's a fourth straight year above $100 million, and more than $796 million since 2019. The Legal Aid Society notes the figure leaves out matters settled before formal litigation. Los Angeles has paid about $384 million across nearly 2,000 police-related claims since September 2019, with civil rights violations, shootings, excessive force, and illegal searches making up 48 percent of the top categories. Philadelphia paid about $60 million in the eighteen months ending September 2024, against a 2013 to 2017 average of $9 million a year, and the driver was case severity rather than case volume.

How often officers are charged, and convicted

Philip Stinson's Police Integrity Research Group at Bowling Green State University tracked on-duty shootings from January 2005 through June 2019. In that period, 104 nonfederal sworn officers were arrested for murder or manslaughter arising from an on-duty shooting, about 6.5 a year. Of 80 concluded cases, 35 ended in conviction and 45 didn't. Of the 35 convictions, only four were for murder.

Set that against roughly a thousand or more fatal police shootings a year and the charging rate runs well under one percent of fatal shootings. Mapping Police Violence, using a broader denominator that includes all killings rather than shootings alone, reports officers charged in 15 incidents in 2025 and a charging rate of roughly four percent, up from about two percent in earlier estimates. The 2025 Police Violence Report puts it under three percent.

Those three framings aren't inconsistent. They use different denominators and different definitions of "charged." Say which one you're using, every time.

Section 1983 outcomes

• Plaintiff trial win rate in constitutional tort cases runs around 30 percent, in both bench and jury trials, against roughly 40 percent for employment cases and over 50 percent for contract and general tort cases. Theodore Eisenberg, "Four Decades of Federal Civil Rights Litigation," 12 Journal of Empirical Legal Studies 1 (2015).
• Civil rights cases settle at 50 to 65 percent, against 70 to 80 percent for non-civil-rights cases. Same source.
• Qualified immunity was raised in about 37.6 percent of cases where available and granted about 17.9 percent of the time, dismissing only 3.2 percent of all cases. Schwartz (2017), 1,183 cases across five federal districts.
• Where immunity was raised and the case reached a jury, plaintiffs won just over 25 percent and defendants nearly 70 percent. Reinert (2018), 211 cases.
• Forum is worth more than most facts. The Fifth Circuit granted immunity in 64 percent of police requests and the Ninth Circuit in 42 percent, per the 2020 Reuters analysis of 529 appellate rulings.

Body-worn cameras

The National Bureau of Economic Research working paper on body-worn cameras (2021) reported a 16.9 percent drop in complaints against officers, which is statistically significant. It also reported a 9.6 percent drop in use of force, which isn't: the confidence interval crosses zero. State that honestly. The claim that cameras reduce use of force is weaker than it's usually asserted to be. The same analysis put the benefit-to-cost ratio at about five to one.

On whether footage changes litigation outcomes, the literature has no answer yet. The case record is suggestive and cuts both ways. Footage produced convictions in several of the matters in Part 13. In others no footage reached the case. The officers never turned the cameras on, the agency sat on the video for two years, or a state release statute kept it from the public. Treat the question as unresolved, and treat the presence or absence of footage as a fact to be investigated rather than assumed.

Attorneys retain me to make a complex sequence of seconds clear to a jury. The work is the same discipline whether the case is a handling violation, a use-of-force decision, or both, and whether the shooter wore a badge.

A good expert does more than display his or her own skill and knowledge. They teach the jury enough about the matter that the jurors reach the same conclusion they did, on their own, behind closed doors.

Three questions I apply to every matter

1. Was the harm foreseeable? Not whether this exact outcome was predicted, but whether the class of outcome was the likely product of the conduct. A person startled awake reaches. A person blinded by a light can't identify who is shouting. A vehicle accelerates. None of that requires prediction. It requires recognition.
2. Was the harm preventable? Name the specific alternatives that existed at the specific moment, and say what each one would have changed. Cover. Distance. Containment. An announcement. A translated command. Thirty seconds of patience. If nothing available would have changed the outcome, say so. That makes most of the defense case.
3. Did the knowledge that would have prevented it exist, and was it available to this shooter? That question separates a standards opinion from a training opinion from a municipal opinion. The published body of practice is one thing. What this agency taught this officer is another, and it's answered from the training file, not from the literature.

Frame-accurate reconstruction

Break the sequence into discrete decision points wherever video exists, whether from store cameras, body-worn cameras, dash cameras, or bystander phones: first information, arrival, first contact, first show of force, each escalation, the shot or other application of force, and the end of the threat. Distance, movement direction, elapsed time, and hand position at each point get measured, not estimated.

Work from the native files with metadata intact, never from a released clip. A clip has a cut point somebody chose. Most systems record a pre-event buffer before activation, and that buffer is often the most informative segment in the case. Synchronize multiple cameras against a common event, and state elapsed time to the hundredth of a second at each point.

Respect what video can't do. Under Scott v. Harris, a court won't discount an opinion that contradicts what the record plainly shows. It will disregard it as a matter of law. The legitimate work is explaining the limits: frame rate, field of view, lens distortion, rolling shutter artifact, low-light behavior, automatic gain, and the difference between the camera's position and the officer's position. A camera on a chest doesn't see what eyes see, and it especially doesn't see what eyes see in the dark.

Audio analysis of the shot string

An audio waveform gives you every shot-to-shot interval across a volley to a few thousandths of a second. It gives you the time from the last command to the first round too. Plumhoff forecloses the argument about round count. It doesn't foreclose the question of when the threat ended, and that measurement answers it.

Match each interval against the wound paths, the trajectory data, and each officer's position at each moment. Where more than one officer fired, the waveform separates the shooters and shows who went first. A contagious-fire analysis rests on that.

Compliance-interval reconstruction

Where a command preceded the force, the question is whether the interval allowed a human being to comply. Answer it with timed trials. Use participants of similar build and circumstance, and measure time to orient and time to first purposeful movement under the conditions that existed: waking from sleep, a light in the eyes, a shouted command, a second language.

Report the mean and the spread, describe the protocol, and state its limits. A Rule 702 challenge lands here, so document the methodology before you run the first trial.

Scene inspection, at the same hour

A scene at two in the afternoon isn't the scene that existed at three in the morning. Measure the light at the exact position. Map the pole lighting and its output, the ambient level inside a vehicle with a flashlight on the driver, the headlight and spotlight positions, and the effect of tinted or wet glass. A daytime photograph gives you none of it.

Ask for permission to inspect at the hour of the incident and, where seasonally relevant, near the date. Where a scene has changed, photograph what changed and document when.

Exemplar comparison

Where the identification of an object is at issue, acquire the same make and model and photograph it beside the weapon it was mistaken for, under the measured scene lighting, from the shooter's position and from the subject's position. That comparison does more to settle a mistaken-object question than any amount of testimony, and it does it whichever way the answer runs.

Vehicle and physical evidence examination

Bullet paths and angles through glass and sheet metal. Glass condition and window position at the time of the shots. Seat position and recline. Ignition state. The position of any object relative to the occupant. Casings and their distribution. Where the vehicle itself is evidence, demand preservation and an inspection. Part 17 covers that demand.

Training and policy review

For a security guard, compare the state licensing status, firearms qualification records, use-of-force policy, and de-escalation training against the state regulatory minimums. Then compare them against industry guidance such as the ASIS International standard for private security officer selection and training.

For an officer, the comparison runs against the department's own use-of-force policy in the version in force on the date of the incident, academy and in-service training records, field training evaluations, simulator and scenario logs, qualification scores and dates, and any prior findings in the officer's file. A gap between the policy on paper and the conduct on video is often the clearest evidence in the case, for either kind of shooter.

Read the policy versions on either side of the incident date. Post-incident policy changes tell you what the department knew was wrong.

Comparative and foreseeability analysis

For the employer or municipal claims, ask what the employer accepted. Was this shooter a foreseeable risk in this role, with this equipment and this training? Prior incidents, complaint history, early-intervention data, staffing decisions, supervisor-to-officer ratio, and hours worked in the days before the event all bear on it. Fatigue is measurable from a schedule and it's rarely asked for.

Demonstratives and testimony

I build timelines, distance diagrams, lighting comparisons, synchronized multi-camera views, and side-by-side comparisons of policy language against video conduct, all of it made to survive a Daubert or Frye challenge and to hold up under cross-examination. That work rests on staying inside the evidence and inside my expertise.

My lane, and the lawyer's

A firearms and use-of-force expert testifies about facts a jury can't supply for itself. A lawyer argues what those facts mean under the law. The line between the two isn't a courtesy. It's the line that decides whether my testimony comes in.

Rule 702 of the Federal Rules of Evidence, as amended effective December 1, 2023, puts the burden on the party offering me. That party has to show the court that it is "more likely than not" that four things are true. My knowledge will help the jury. My testimony rests on sufficient facts. It comes from reliable principles and methods. And my opinion "reflects a reliable application" of those methods to the facts of this case. The last clause is new wording, and it's where these motions now land. It isn't enough that a method is sound in general. The opinion has to show the method applied here.

Rule 704(a) says an opinion isn't objectionable just because it embraces an ultimate issue. That rule is narrower than experts want it to be. Berry v. City of Detroit, 25 F.3d 1342 (6th Cir. 1994), reads it this way: the ultimate issue embraced has to be a factual one, and "it is the responsibility of the court, not testifying witnesses, to define legal terms." The court in Berry drew the line at exactly the place I have to respect. A police-practices expert could testify that discipline in the department was lax and describe what lax discipline produces. He couldn't testify that lax discipline made the city deliberately indifferent. The first is an observation. The second is a legal conclusion wearing an expert's coat. Specht v. Jensen, 853 F.2d 805 (10th Cir. 1988) (en banc), says the same thing more bluntly: a witness can't give an opinion on a question of law, and it isn't for witnesses to instruct the jury on the applicable principles.

Rule 704(b) adds a hard bar in criminal cases. An expert must not state an opinion about whether a defendant had a mental state that is an element of the crime or of a defense. So in a criminal prosecution of a shooter, I can describe what a trained officer perceives at a given distance under a given light level in a given number of milliseconds. I can't say what this officer actually believed.

Read the right-hand column as a list of sentences I don't write. Hygh v. Jacobs, 961 F.2d 359 (2d Cir. 1992), is the case that names them. There an expert testified that an officer's force was not "justified under the circumstances." That his use of deadly physical force was not "warranted under the circumstances." That his conduct was "totally improper." The Second Circuit held all of it was error. That testimony, the court said, told the jury what result to reach.

Two notes on how to use Hygh. First, the liability judgment was affirmed anyway on other grounds, so the case isn't a warning that this language loses trials. It's a warning that the language draws a motion and gives the other side a clean cross-examination. Second, the phrases the court struck are close cousins of ordinary words I do use. "Departed from accepted practice" is a factual comparison against a documented standard. "Not warranted under the circumstances" is a verdict. The difference is whether I'm measuring against something in the record or announcing a conclusion.

Why experience qualifies me, and where experience stops

I have no doctorate. My foundation is more than thirty-five years of handling, loading, discharging, and training others on these weapons, on sets and on ranges, plus the records in this case. That foundation is enough, and Berry explains why. A beekeeper with no scientific training could properly testify that bumblebees take off into the wind, because he has watched a great many bumblebees. He knows nothing more about flight principles than the jurors do. He has just seen more bees. The foundation isn't the degree. It's the firsthand observation, laid out so the court can see it.

One caution on citing Berry for this: the court used the beekeeper to explain the principle and then found the expert in front of it hadn't met it. Cite the passage for the rule, not as a holding that an experience-based expert walks in free.

Kopf v. Skyrm, 993 F.2d 374 (4th Cir. 1993), is the affirmative authority, and it's weapon-specific. The trial court had excluded a former police chief and a canine trainer outright. The Fourth Circuit reversed, and said the police chief "should clearly have been permitted to testify as to the prevailing standard of conduct for the use of slapjacks, even if he had been precluded from giving an opinion on the ultimate issue." That sentence is the whole architecture of my testimony. Prevailing standards for a specific tool are expert territory. The ultimate issue isn't, and losing it costs me nothing, because the standards testimony is what the jury actually needs.

Experience alone won't save an opinion that skips the work. Slatowski v. Sig Sauer, Inc., 148 F.4th 132 (3d Cir. 2025), decided August 1, 2025, affirmed the exclusion of two experienced experts in a firearm design case because neither had simulated the conditions of the shooting. Their theory was that the pistol could fire more easily. They hadn't tested whether it did, under these conditions, in this event. Read the case in full before you cite it, because the same opinion vacated summary judgment for the manufacturer and sent the case to trial on the remaining evidence. The lesson isn't that experts lose. It's that credentials don't substitute for testing tied to this incident, which is why the scene work, the exemplar comparison, and the timed reconstruction described above aren't optional.

Policy isn't the legal standard, and I say so

Departmental policy is evidence. It isn't the test. United States v. Brown, 871 F.3d 532 (7th Cir. 2017), puts it plainly: "the excessive-force inquiry is governed by constitutional principles, not police-department regulations," and departmental compliance is "neither sufficient nor necessary to satisfy the Fourth Amendment's reasonableness requirement." The same opinion warns that "evidence of purely localized police procedure is less likely to be helpful than nationally or widely used policy," because the standard contemplates a reasonable officer, not a reasonable person.

That cuts both ways, and I say it in the report whichever side retained me. For a plaintiff: a policy violation is powerful evidence of a departure from accepted practice, and it isn't by itself a constitutional violation. For a defense: policy compliance is real evidence of reasonable conduct, and it isn't a defense on its own. Anchor the practice opinion to national standards and published training doctrine, then use the local policy as corroboration.

A worked example

A case turns on whether a device was a firearm within the meaning of a statute. I can state what the device is, how it operates, what it discharges, at what velocity, with what energy at the measured distance, and how it compares to weapons I've handled and tested. I stop there. Whether that description satisfies the statutory definition is a question of law, and counsel briefs it using my description. If I answer it myself I've done the lawyer's job badly and my own job not at all, and I've handed the other side a motion.

The same discipline covers state statutes that turn on necessity. An opinion that force was unnecessary is a state-law opinion in the states that wrote necessity into their statutes. It isn't a Fourth Amendment opinion, because the Fourth Amendment asks about reasonableness. I write one opinion per standard, and I name the standard in the opinion's first line.

What to write, and what to leave to the jury

Segregate the opinions by the standard each one addresses. One opinion on agency policy. One on the state statutory standard, where the state has one. One on generally accepted practices. One on the physical reconstruction. Then stop. The jury decides whether the conduct was unreasonable, whether the agency was indifferent, and whether anyone should pay. My job is to give them the facts they can't get anywhere else and to leave those three questions untouched.

I bill for my time, not my opinion.

Most of what decides a use-of-force case is perishable. Third-party surveillance overwrites on a thirty-day loop. Vehicles get repaired or sold. Digital evidence retention categories expire. Scenes get relit and repaved.

Send the preservation letter before anything else, and send it wider than the defendant agency. The list runs in the order the work runs. It applies to a police matter, a security matter, or a civilian shooting, with the obvious substitutions. Ask for what you need in the order you need it, and ask what preservation demand was already sent, by whom, and when.

Video, audio, and the timeline

The physical evidence

Training and the individuals

Policy

The investigation and any declination

Communications

Pattern and practice, for the municipal claim

The subject's side

The same record supports two reports. The evidence picks which one. Build both and keep the one that survives.

Run the record against both columns before you commit to a theory, and run it again after discovery closes.

The case for a justified use of force

Every item below is a finding, not an argument. Each one has a document or a measurement behind it.

The case for an unjustified use of force

The five ways an opinion in this field gets destroyed

1. Opining against plain video. Scott v. Harris means an account blatantly contradicted by the record gets disregarded as a matter of law. Explain what the camera couldn't capture. Never claim it captured something it didn't.
2. Sliding between standards. Unnecessary is a state-law word in some states and a meaningless one federally. Unreasonable is a constitutional term the jury owns. Deliberately indifferent is the jury's too. Say which standard each opinion addresses and stay inside it.
3. Treating best practice as the constitutional floor. It isn't. The defense will make you concede that, and a late concession looks like retreat. Concede it first and reframe: national standards evidence what the agency knew about the risk.
4. Advancing the reserved question. "He created the danger, so the shooting was unreasonable" is the formulation Mendez rejected and the Supreme Court has twice declined to adopt. Part 8 gives three doctrinal homes for the same evidence that don't require it.
5. Opining without testing these conditions. Experience qualifies the witness. It doesn't supply the method. Slatowski v. Sig Sauer, 148 F.4th 132 (3d Cir. 2025), affirmed the exclusion of two experienced firearms experts who never simulated the conditions of the shooting. Run the scene at the same hour, time the sequence, fire the exemplar, and put the protocol in the report. Part 16 sets out what the rules require and what they forbid.

Store managers, security companies, property owners, and agency trainers read these matters after the fact. They want to know what should have happened instead. It's also the plainest statement of the standard a jury will be asked to apply.

The private-actor decision ladder

Measure every escalation in a detention against the threat presented at that moment, not against the decision to intervene. A shoplifting suspect who hasn't displayed a weapon and isn't attacking anyone doesn't justify skipping rungs on this ladder.

The approach ladder for a reported armed subject

This one is for officers and the people who train them. This sequence keeps the Part 8 fact patterns from happening. A plaintiff's expert will hold the video against it.

Common questions

Authority, public reporting, and records consulted for the figures and matters in this guide. Verify the current status of any case, statute, or active matter before relying on it in a filing.

Supreme Court authority

1. Graham v. Connor490 U.S. 386 (1989). The Fourth Amendment objective-reasonableness standard for police use of force, the three factors, the bar on hindsight, and the removal of intent from the analysis.
2. Tennessee v. Garner471 U.S. 1 (1985). Deadly force against a fleeing, non-dangerous suspect is an unreasonable seizure. The warning requirement, conditioned on feasibility.
3. Barnes v. Felix605 U.S. 73 (2025). Decided May 15, 2025, unanimously. Rejected the moment-of-threat doctrine; the totality inquiry has no time limit. Expressly reserved whether an officer's own creation of a dangerous situation factors into reasonableness. On remand the Fifth Circuit affirmed judgment for the officer, No. 22-20519 (5th Cir. Sept. 18, 2025); rehearing denied November 14, 2025; a second petition for certiorari denied April 27, 2026. No published federal reporter citation for the remand opinion was located; verify before citing.
4. County of Los Angeles v. Mendez581 U.S. 420 (2017). Rejected the provocation rule, and preserved recovery for damages proximately caused by a separate Fourth Amendment violation. On remand, Mendez v. County of Los Angeles, 897 F.3d 1067 (9th Cir. 2018), affirmed full damages on the corrected proximate-cause theory.
5. Scott v. Harris550 U.S. 372 (2007). Terminating a dangerous pursuit; the video rule; Garner isn't a separate checklist.
6. Plumhoff v. Rickard572 U.S. 765 (2014). Officers justified in firing need not stop shooting until the threat has ended.
7. Torres v. Madrid592 U.S. 306 (2021). Physical force applied with intent to restrain is a seizure even where the person isn't subdued.
8. Lombardo v. City of St. Louis594 U.S. 464 (2021). Prone restraint; duration, weight, position, restraint status, and the agency's own training warnings are constitutionally material.
9. Kingsley v. Hendrickson576 U.S. 389 (2015). Pretrial detainees need show only that the force was objectively unreasonable. The Kingsley factors.
10. County of Sacramento v. Lewis523 U.S. 833 (1998). No seizure in an accidental pursuit collision; the shocks-the-conscience standard and the deliberation sliding scale.
11. City of Tahlequah v. Bond595 U.S. 9 (2021). Immunity granted; expressly declined to decide whether recklessly creating a situation requiring deadly force can itself violate the Fourth Amendment.
12. White v. Pauly580 U.S. 73 (2017). A late-arriving officer may assume proper procedures were already followed. Officer-by-officer analysis is required.
13. Rivas-Villegas v. Cortesluna595 U.S. 1 (2021), and Kisela v. Hughes, 584 U.S. 100 (2018), and Mullenix v. Luna, 577 U.S. 7 (2015). The specificity requirement for clearly established law.
14. Zorn v. Linton607 U.S. (2026), No. 25-297, decided March 23, 2026. Per curiam summary reversal granting qualified immunity; principles stated generally don't suffice; a prior appellate denial of summary judgment doesn't clearly establish law. Sotomayor, J., dissenting, joined by Kagan and Jackson, JJ.
15. Case v. Montana607 U.S. (2026), No. 24-624, decided January 14, 2026. Warrantless emergency-aid entry requires an objectively reasonable basis for believing an occupant is seriously injured or imminently threatened, not probable cause. Sotomayor, J., concurring, on police presence escalating crisis encounters.
16. Monell, Canton, Connick, Pembaur, Praprotnik, Bryan CountyMonell v. Department of Social Services, 436 U.S. 658 (1978); City of Canton v. Harris, 489 U.S. 378 (1989); Connick v. Thompson, 563 U.S. 51 (2011); Pembaur v. City of Cincinnati, 475 U.S. 469 (1986); City of St. Louis v. Praprotnik, 485 U.S. 112 (1988); Board of County Commissioners of Bryan County v. Brown, 520 U.S. 397 (1997). Municipal liability, failure to train, ratification, and negligent hiring.
17. Owen and Fact ConcertsOwen v. City of Independence, 445 U.S. 622 (1980), no qualified immunity for municipalities; City of Newport v. Fact Concerts, 453 U.S. 247 (1981), no punitive damages against municipalities.
18. Saucier, Pearson, al-Kidd, Hope, TaylorSaucier v. Katz, 533 U.S. 194 (2001); Pearson v. Callahan, 555 U.S. 223 (2009); Ashcroft v. al-Kidd, 563 U.S. 731 (2011); Hope v. Pelzer, 536 U.S. 730 (2002); Taylor v. Riojas, 592 U.S. 7 (2020). The qualified immunity framework and the narrow obvious-case exception.
19. Bivens and its narrowingBivens v. Six Unknown Named Agents, 403 U.S. 388 (1971); Egbert v. Boule, 596 U.S. 482 (2022); Goldey v. Fields, 606 U.S. (2025), No. 24-809, decided June 30, 2025. The federal-officer damages remedy is largely closed.
20. Richardson, Filarsky, GriffinRichardson v. McKnight, 521 U.S. 399 (1997); Filarsky v. Delia, 566 U.S. 377 (2012); Griffin v. Maryland, 378 U.S. 130 (1964). State action and immunity for private actors performing government functions.

Courts of appeals and state courts

1. Officer-created jeopardy, Tenth CircuitSevier v. City of Lawrence, 60 F.3d 695 (10th Cir. 1995); Allen v. Muskogee, 119 F.3d 837 (10th Cir. 1997); Hastings v. Barnes, 252 F. App'x 197 (10th Cir. 2007); Estate of Ceballos v. Husk, 919 F.3d 1204 (10th Cir. 2019); Flores v. Henderson, 101 F.4th 1185 (10th Cir. 2024).
2. Officer-created jeopardy, other circuitsEstate of Starks v. Enyart, 5 F.3d 230 (7th Cir. 1993); Abraham v. Raso, 183 F.3d 279 (3d Cir. 1999); Vos v. City of Newport Beach, 892 F.3d 1024 (9th Cir. 2018); Young v. City of Providence, 404 F.3d 4 (1st Cir. 2005); Young v. Borders, 850 F.3d 1274 (11th Cir. 2017), rehearing en banc denied.
3. Segmenting and the later volleyDickerson v. McClellan, 101 F.3d 1151 (6th Cir. 1996); Claybrook v. Birchwell, 274 F.3d 1098 (6th Cir. 2001); Waterman v. Batton, 393 F.3d 471 (4th Cir. 2005).
4. Narrow-frame authority, abrogated on timing by BarnesSalim v. Proulx, 93 F.3d 86 (2d Cir. 1996); Schulz v. Long, 44 F.3d 643 (8th Cir. 1995); Elliott v. Leavitt, 99 F.3d 640 (4th Cir. 1996); Rockwell v. Brown, 664 F.3d 985 (5th Cir. 2011).
5. State negligence reaching pre-shooting conductHayes v. County of San Diego, 57 Cal. 4th 622 (2013); Tabares v. City of Huntington Beach, 988 F.3d 1119 (9th Cir. 2021).
6. Bystanders and the seizure requirementBrower v. County of Inyo, 489 U.S. 593 (1989); Rucker v. Harford County, 946 F.2d 278 (4th Cir. 1991); Landol-Rivera v. Cruz Cosme, 906 F.2d 791 (1st Cir. 1990); Medeiros v. O'Connell, 150 F.3d 164 (2d Cir. 1998); Childress v. City of Arapaho, 210 F.3d 1154 (10th Cir. 2000); Claybrook v. Birchwell, 199 F.3d 350 (6th Cir. 2000), which is the bystander holding rather than 274 F.3d 1098; Hawatmeh v. City of Henderson, No. 24-6146 (9th Cir. Nov. 7, 2025). The other side: Vaughan v. Cox, 343 F.3d 1323 (11th Cir. 2003), the 2001 opinion at 264 F.3d 1027 having been vacated; Nelson v. City of Davis, 685 F.3d 867 (9th Cir. 2012); Villanueva v. California, 986 F.3d 1158 (9th Cir. 2021); Kilnapp v. City of Cleveland, No. 25-3149 (6th Cir. Feb. 18, 2026), published. Mistaken identity as a seizure: Milstead v. Kibler, 243 F.3d 157 (4th Cir. 2001). Compare Corbitt v. Vickers, 929 F.3d 1304 (11th Cir. 2019).
7. Cunningham v. Baltimore County487 Md. 282 (2024), cert. denied, No. 24-578 (Jan. 27, 2025). The mother's estate recovered under the Fourth Amendment as the target. The child struck by the same round after it passed through her did not, and the verdict on his behalf was taken away on qualified immunity.
8. Martin v. United States605 U.S. 395 (2025). No Supremacy Clause defense to a Federal Tort Claims Act suit, and the law enforcement proviso in 28 United States Code section 2680(h) doesn't override the discretionary function exception in section 2680(a). See also Brownback v. King, 592 U.S. 209 (2021), on the judgment bar.
9. Transferred justificationRestatement (Third) of Torts: Intentional Torts to Persons section 26(c): the privilege defeats the bystander's intentional tort claim, and liability is then determined by negligence principles. Criminal: Commonwealth v. Fowlin, 551 Pa. 414, 710 A.2d 1130 (1998); Commonwealth v. Santana-Rodriguez, SJC-13753 (Mass. Oct. 14, 2025); Model Penal Code section 3.09(3).
10. Plakas v. Drinski19 F.3d 1143 (7th Cir. 1994). No constitutional duty to exhaust non-deadly alternatives before otherwise-justified deadly force.
11. Private-actor authorityCervantez v. J.C. Penney Co., 24 Cal. 3d 579 (1979); Katko v. Briney, 183 N.W.2d 657 (Iowa 1971); People v. Ceballos, 12 Cal. 3d 470 (1974).

State tort, culpability, and damages authority

1. South v. Maryland59 U.S. 396 (1855). Origin of the public duty doctrine. A duty owed to the public at large is owed to no one in particular.
2. Anderson v. Massillon134 Ohio St. 3d 380 (2012). Willful, wanton, and reckless describe different degrees of care and aren't interchangeable. Ohio ranks wanton misconduct above reckless conduct.
3. Ziarko v. Soo Line Railroad161 Ill. 2d 267 (1994). Willful and wanton as a hybrid category running from a shade above negligence to a shade below intentional wrongdoing.
4. The McHaffie ruleMcHaffie v. Bunch, 891 S.W.2d 822 (Mo. 1995), and the states following it: an admission of respondeat superior bars direct negligent hiring, training, and supervision claims against the employer. Verify the rule in the forum, and check whether a punitive damages claim defeats it there.
5. Restatement (Second) of Torts section 500Reckless disregard of safety. The actor knows or has reason to know of facts creating a high degree of risk and deliberately proceeds, or fails to realize the risk where it would be apparent to a reasonable person.
6. Restatement (Second) of Torts section 120AShopkeeper's privilege. Reasonable grounds, reasonable time, reasonable manner, non-deadly force only.
7. Restatement (Second) of Torts section 909Complicity rule for punitive damages against a principal, including subsection (b), the negligent-hiring route.
8. Smith v. Wade461 U.S. 30 (1983). Punitive damages against an individual under Section 1983 on evil motive or intent, or reckless or callous indifference. The test is disjunctive.
9. City of Newport v. Fact Concerts453 U.S. 247 (1981). No punitive damages against a municipality under Section 1983.
10. Punitive damages, constitutional limitsBMW of North America v. Gore, 517 U.S. 559 (1996); State Farm v. Campbell, 538 U.S. 408 (2003). Due process guideposts on punitive awards: reprehensibility, ratio, and comparable civil penalties.

Expert testimony and admissibility

1. Federal Rule of Evidence 702As amended effective December 1, 2023. The proponent must demonstrate to the court that it is more likely than not that each requirement is met, and subsection (d) now asks whether the opinion "reflects a reliable application" of the principles and methods to the facts of the case.
2. Federal Rule of Evidence 704Subsection (a): an opinion isn't objectionable just because it embraces an ultimate issue. Subsection (b): in a criminal case an expert must not opine on whether the defendant had a mental state that is an element of the charge or of a defense.
3. Berry v. City of Detroit25 F.3d 1342 (6th Cir. 1994). The ultimate issue embraced must be a factual one, and it is the court's responsibility, not the witness's, to define legal terms. Contains the bumblebee and beekeeper passage on experience-based qualification. The court stated the principle and then found the expert before it hadn't satisfied it, so cite the passage for the rule, not as a holding of qualification.
4. Kopf v. Skyrm993 F.2d 374 (4th Cir. 1993). Reversal of the wholesale exclusion of a police-practices expert and a canine trainer. The expert should have been permitted to testify to the prevailing standard of conduct for the use of a slapjack "even if he had been precluded from giving an opinion on the ultimate issue." The best affirmative authority for weapon-specific standards testimony.
5. Hygh v. Jacobs961 F.2d 359 (2d Cir. 1992). Error to admit expert testimony that force was not "justified under the circumstances," not "warranted under the circumstances," and "totally improper." Such testimony told the jury what result to reach. The excessive-force judgment was nevertheless affirmed, so read the case as a warning about motions and cross-examination, not about outcomes.
6. Specht v. Jensen853 F.2d 805 (10th Cir. 1988) (en banc). A witness can't give an opinion on a question of law, and it isn't for witnesses to instruct the jury on the applicable principles.
7. United States v. Brown871 F.3d 532 (7th Cir. 2017). The excessive-force inquiry is governed by constitutional principles, not police-department regulations; departmental compliance is neither sufficient nor necessary under the Fourth Amendment. Nationally or widely used policy is more likely to help the jury than purely localized procedure.
8. Slatowski v. Sig Sauer148 F.4th 132 (3d Cir. 2025). Decided August 1, 2025. Exclusion of two experienced experts affirmed because neither simulated the conditions of the shooting. Note that the same opinion vacated summary judgment for the manufacturer and remanded for trial.
9. The gatekeeping trilogyDaubert v. Merrell Dow Pharmaceuticals, 509 U.S. 579 (1993); General Electric v. Joiner, 522 U.S. 136 (1997); Kumho Tire v. Carmichael, 526 U.S. 137 (1999). The federal gatekeeping trilogy. Kumho extends gatekeeping to technical and experience-based testimony. A minority of states still apply the general-acceptance test of Frye v. United States, 293 F. 1013 (D.C. Cir. 1923); confirm the forum's standard before drafting.

Statutes and codes cited

1. FederalTitle 42, United States Code, section 1983.
2. Strict-standard use-of-force statutesCalifornia Penal Code sections 835a and 13519.10; Revised Code of Washington sections 9A.16.040 and 10.120.020; Colorado Revised Statutes section 18-1-707; Connecticut General Statutes section 53a-22; 720 Illinois Compiled Statutes 5/7-5; Maryland Public Safety section 3-524; Minnesota Statutes section 609.066; Virginia Code Title 19.2, Chapter 7.1; Oregon Revised Statutes section 161.242; Nevada Revised Statutes section 171.1455; Utah Code section 76-2-404; Massachusetts Acts of 2020, Chapter 253.
3. State causes of action barring qualified immunityColorado Revised Statutes section 13-21-131; New Mexico Statutes Annotated sections 41-4A-1 and following; Connecticut General Statutes section 52-571k (preserving a good-faith defense); New York City Administrative Code section 8-802 and following.
4. Private-actor force and arrestTexas Penal Code sections 9.32, 9.41, 9.42, 9.43, and 9.51; California Penal Code sections 197 and 837; New York Penal Law section 35.30(4); Mississippi Code section 97-3-15; Georgia House Bill 479 (2021), repealing Official Code of Georgia Annotated sections 17-4-60 through 17-4-62 and enacting section 17-4-80; Restatement (Second) of Torts sections 79, 85, and 120A.

Standards, training, and research

1. Police Executive Research ForumGuiding Principles on Use of Force (2016). The sanctity of human life, proportionality above the constitutional minimum, the critical decision-making model, tactical repositioning, distance plus cover equals time, the duty to intervene, and Guiding Principle 8 on shooting at moving vehicles.
2. Integrating Communications, Assessment, and TacticsEngel, Corsaro, Isaza, and McManus, evaluation of the Louisville Metro Police Department implementation, Criminology and Public Policy, volume 21, issue 2 (2022). A stepped-wedge randomized controlled trial reporting reductions of 28.1 percent in use-of-force incidents, 26.3 percent in community-member injuries, and 36.0 percent in officer injuries.
3. National Consensus Policy and Discussion Paper on Use of ForceFirst released 2017, developed by eleven law enforcement organizations including the International Association of Chiefs of Police. Retains the objective-reasonableness standard rather than a necessity standard, while adding de-escalation, duty to intervene, warning where possible, and restrictions on shooting at moving vehicles. The Police Executive Research Forum splits from this model on the necessity question. Know that before a deposition, because each side will cite the model that helps it.
4. Reaction timeBlair, Pollock, Montague, Nichols, Curnutt, and Burns, "Reasonableness and Reaction Time," Police Quarterly, volume 14, issue 4 (2011). The article reports that officers generally couldn't beat a subject who had chosen to fire. The frequently cited mean values of roughly four tenths of a second for officer reaction and subject firing time appear in the article body; verify against the published article before quoting a figure.
5. Hit ratesBernard D. Rostker and others, "Evaluation of the New York City Police Department Firearm Training and Firearm-Discharge Review Process," RAND Corporation MG-717 (2008): an 18 percent hit rate in gunfights and 30 percent where fire wasn't returned, 1998 through 2006. Christopher M. Donner and Nicole Popovich, "Hitting (or missing) the mark: An examination of police shooting accuracy in officer-involved shooting incidents," Policing: An International Journal (2018): a 35 percent bullet-level hit rate across Dallas Police Department shootings from 2003 through 2017, with a hit in only 54 percent of incidents.
6. Overpenetration and ricochetUrey W. Patrick, "Handgun Wounding Factors and Effectiveness," Federal Bureau of Investigation Firearms Training Unit (1989), the source of the 12 to 18 inch penetration window. The Bureau's ammunition test protocol, including the wallboard event, two pieces of half-inch gypsum board set three and a half inches apart with the gelatin block 18 inches behind. Hlavaty and others, "Examination of Ricochet Gunshot Wounds from Commonly Encountered Surfaces," American Journal of Forensic Medicine and Pathology, volume 37 (2016).
7. Federal use-of-force policyDepartment of Homeland Security Policy Statement 044-05, "Update to the Department Policy on the Use of Force," Revision 1, February 6, 2023, including the requirement that an officer consider the hazards posed to innocent bystanders. Secret Service authority at 18 United States Code section 3056. No Secret Service directive of its own is public.
8. The 21-foot constructDennis Tueller, "How Close is Too Close?", S.W.A.T. Magazine, March 1983. A training demonstration of the action-and-reaction problem, widely misused as a legal bright line.
9. New York City Police Department firearms dataAnnual use-of-force and firearms discharge reporting. The 2024 edition reports 28 intentional adversarial-conflict discharge incidents, 150 rounds fired, 22 subjects struck, and four bystanders injured across three separate incidents. The department reports an objective completion rate rather than a hit ratio; don't present the one as the other.
10. Contagious fireWhite and Klinger, "Contagious Fire? An Empirical Assessment of the Problem of Multi-shooter, Multi-shot Deadly Force Incidents in Police Work," Criminal Justice and Behavior (2012); DeCarlo, Dlugolenski, and Myers, an experimental test of the contagious fire thesis in policing, Journal of Criminal Justice (2024).
11. Weapon confusionHoward E. Williams, "Weapon Confusion: TASER CEWs, Firearms, and Human Error Theories," Criminal Justice Review (2022); Jeffrey A. Martin, "Applied Human Error Theory: A Police Taser-Confusion Shooting Case Study," Proceedings of the Human Factors and Ergonomics Society (2016).
12. Academy trainingBureau of Justice Statistics, State and Local Law Enforcement Training Academies and Recruits, 2022.
13. ASIS InternationalPrivate Security Officer Selection and Training standard, referenced as an industry benchmark for guard training review.
14. Body-worn camerasWilliams, Weil, Rasich, Ludwig, Chang, and Egrari, "Body-Worn Cameras in Policing: Benefits and Costs," National Bureau of Economic Research Working Paper 28622 (2021).

Data and empirical studies

1. Mapping Police Violence2025 year-end report and the 2025 Police Violence Report, Campaign Zero. Counts all deaths from any application of force by any officer, on or off duty.
2. Washington Post, Fatal ForceFatal police shootings only, on duty, 2015 through 2024. Discontinued in January 2025. Say so in any current citation.
3. The Lancet"Fatal police violence by race and state in the USA, 1980 to 2019: a network meta-regression," October 2021. Estimated that 55.5 percent of deaths from police violence were misclassified or unreported in federal vital statistics.
4. Settlement costsWashington Post investigation, March 2022, on repeated police misconduct across the 25 largest departments; current city-level reporting for Chicago, New York City, Los Angeles, and Philadelphia.
5. IndemnificationJoanna C. Schwartz, "Police Indemnification," 89 New York University Law Review 885 (2014).
6. Qualified immunity, empiricalJoanna C. Schwartz, "How Qualified Immunity Fails," 127 Yale Law Journal 2 (2017); Alex Reinert, "Qualified Immunity at Trial," 93 Notre Dame Law Review 2065 (2018); Reuters, "Shielded by the Law" investigation (2020).
7. Criminal charging of officersPhilip M. Stinson, Police Integrity Research Group, Bowling Green State University, "On-Duty Shootings: Police Officers Charged with Murder or Manslaughter," covering January 2005 through June 2019.
8. Civil rights litigation outcomesTheodore Eisenberg, "Four Decades of Federal Civil Rights Litigation," 12 Journal of Empirical Legal Studies 1 (2015).
9. Policy adoptionStanford Center for Racial Justice survey of use-of-force policies across the hundred largest cities; Brennan Center, State Policing Reforms Since George Floyd's Murder.
10. Negligent security litigation statisticsPublished study of 1,086 reported cases from 1992 to 2001, "Major Developments in Premises Security Liability III," by Norman D. Bates, Liability Consultants, Inc.

Notes on figures and citations that need checking

• No published federal reporter citation was located for the Fifth Circuit's September 18, 2025 opinion in Barnes v. Felix on remand. Cite it by docket number and date until the reporter cite is confirmed.
• I couldn't confirm the parallel Supreme Court Reporter citation for Barnes v. Felix from a primary source. The United States Reports citation, 605 U.S. 73, is confirmed.
• The dissents to the denial of rehearing en banc in Young v. Borders are widely quoted. I couldn't verify that quotation in the reporter text, so it doesn't appear here. The substance of those dissents is described rather than quoted.
• Published stand-your-ground and duty-to-retreat compilations disagree at the margins. No fifty-state table is printed here.
• Private security training-hour figures in general circulation come from commercial compliance vendors rather than state regulators. Verify against the state licensing authority.
• State tort claims act notice deadlines and damage caps move frequently. Verify each against the current statute.
• No defensible published national median exists for Section 1983 excessive force settlements or verdicts, and none is printed here.
• Singleton v. Casanova, No. 22-50327 (5th Cir. June 10, 2024), is not designated for publication and no federal reporter citation was confirmed. Cite it by docket number, and note that it left the stray-round question unresolved rather than deciding it.
• The legal basis of the October 17, 2025 Denver bystander verdict isn't stated in public reporting. The case was tried in state court against the officer individually with no municipal defendant, which is the signature of Colorado's state cause of action, but pull the complaint before saying so.
• Civil outcomes for the nine bystanders wounded outside the Empire State Building in August 2012 don't appear in any accessible public record. No settlement figure is printed here.
• The settlement in the 2015 Marksville, Louisiana matter is sealed by federal order. Any figure in circulation is unsourced.
• Several matters described in Part 13 were unresolved as this edition went out, including active civil litigation, pending appeals, and pending sentencings. Check the current status of any of them.

Authored works

Wolf Safety Series, by Steve Wolf: Firearms Safety On Set, The Smart Citizen's Guide to Concealed Carry, and Firearm Product-Defect Litigation. Part 3 takes its four conditions for deadly force and its continuum from the framework the series teaches throughout.

Contact

Steve Wolf  |  wolf.steve@gmail.com  |  (512) 653-9653  |  SteveWolfExpertWitness.com

Expert Witness in Human Factors, Firearms, and On-Set Safety

Steve Wolf is a nationwide firearms and safety expert witness, retained by plaintiff and defense counsel in civil and criminal matters. He has spent more than thirty-five years in hands-on work across film special effects and firearms handling, spanning 77 feature films as a special effects coordinator. He has built a strong trial record across plaintiff and defense retentions, including a $66.5 million New Mexico verdict that ranks among the largest personal injury verdicts in the state's history. His cases span shootings, gun ranges, and unintentional discharge; use of force by police, security, and armed civilians; pyrotechnics, fire, and explosions; wildfire and property loss; stunts, rigging, and aerial work; on-set and theatrical incidents; and climbing gyms and ropes courses.

Wolf's command of on-set firearm safety put him at the center of the public analysis of the "Rust" shooting. That 2021 incident, on the New Mexico set of the Alec Baldwin film, killed cinematographer Halyna Hutchins and injured director Joel Souza. Engaged to examine the incident, he laid out how a single-action revolver discharges, why trigger-finger discipline is decisive, and where the production's safety practices broke down. CNN, Fox News, and CBS sought his assessment. It drew on his earlier work investigating movie-set shootings, among them the 1993 death of Brandon Lee.

Attorneys retain Wolf because he makes complex cases clear to a jury. He pinpoints the preventable failure, explains causation in plain language, ties conduct to accepted safety and use-of-force standards, builds demonstratives jurors remember, and holds his ground under cross-examination. The goal isn't to display his own knowledge. It's to teach the jury enough that they reach the same conclusion on their own, behind closed doors.

His expertise rests on decades in film special effects and firearms work, 11 patents, a growing shelf of published professional reference books, and a long record of firearms instruction and on-set safety supervision. He is the author of the Wolf Safety Series, including Firearms Safety On Set, The Smart Citizen's Guide to Concealed Carry, and Firearm Product-Defect Litigation.

Steve Wolf  |  wolf.steve@gmail.com  |  (512) 653-9653  |  SteveWolfExpertWitness.com