The Yale Law Journal


Backdoor Municipal Immunity

14 Oct 2022

abstract. More than forty years ago, in Owen v. City of Independence, the Supreme Court held that local governments are not entitled to the protections of qualified immunity. Yet four federal circuits have concluded that granting an officer qualified immunity dooms a failure-to-train claim against their employer because local governments cannot train officers about law that is not “clearly established.” In this Essay, I argue that these circuits’ conflation of qualified immunity and municipal liability—what I call backdoor municipal immunity—misunderstands the role court decisions actually play in police policies and training and undermines the deterrence and compensation rationales underpinning the Court’s decision in Owen.


When local government officials violate people’s rights, a civil rights lawsuit filed under 42 U.S.C. § 1983 is often the best available means of seeking accountability and justice.1 The Supreme Court has long recognized the importance of Section 1983 suits to civil rights enforcement. As the Court observed in 1980, in Owen v. City of Independence, Section 1983 suits are “a vital component of any scheme for vindicating cherished constitutional guarantees.”2 But over the past several decades, the Supreme Court has made it increasingly difficult for plaintiffs to prevail in Section 1983 suits, even when their constitutional rights have been violated. Among the most significant doctrinal challenges for plaintiffs to overcome are qualified immunity and municipal liability.

Qualified immunity shields individual officers from damages liability so long as they have not violated “clearly established law.”3 The Supreme Court’s definition of what constitutes clearly established law has narrowed significantly in recent years. The Court has instructed lower courts that rights cannot be clearly established at “a high level of generality.”4 Instead, “‘whether the violative nature of particular conduct is clearly established’ . . . ’must be undertaken in light of the specific context of the case, not as a broad general proposition.’”5 That has come to mean, in practice, that a plaintiff can defeat a qualified immunity motion only if they can find a prior court decision that has held nearly identical conduct to be unconstitutional.6

Local governments can also be sued under Section 1983 when their officers violate the Constitution, and governments cannot claim the protections of qualified immunity.7 But the standard for proving municipal liability under Section 1983 imposes different—and equally onerous—burdens. As the Supreme Court made clear in Monell v. Department of Social Services, a municipality can only be held liable under Section 1983 for the constitutional violations of its employees if the municipality had a policy, practice, or custom that caused the constitutional violation.8 One common way in which plaintiffs seek to establish Monell liability is by showing that the local government failed to train its officers adequately.9 The Supreme Court has held that in order to succeed on a “failure-to-train” claim, the plaintiff must show that policymakers were “deliberately indifferent” to the need for better training—either because the need for that training was obvious or because policymakers knew of past similar unconstitutional conduct by their officers yet failed to take proper steps to prevent such misconduct from recurring.10 But the Supreme Court has narrowly defined the types of training needs that are obvious and has required evidence of nearly identical past misconduct to prove policymakers’ deliberate indifference.11

The Supreme Court’s qualified immunity and Monell doctrines make it exceedingly challenging for plaintiffs to prevail in claims against officers and local governments under Section 1983. As Fred Smith has observed, the two doctrines combine to create a de facto local sovereign immunity.12

But some courts have taken things one step further, merging the protections of qualified immunity and Monell. These courts have held that when an officer is found to be entitled to qualified immunity because the right at issue was not “clearly established,” then the municipality cannot have been deliberately indifferent to the need to train its officers about that right.13

This Essay argues against this conflation of qualified immunity and local-government liability—what I refer to as backdoor municipal immunity—because it misunderstands how local governments train their officers about the constitutional limits of their power, subverts the spirit of the Supreme Court’s 1980 ruling in Owen that municipalities should not receive qualified immunity, and frustrates Section 1983 plaintiffs’ pursuits of justice in jurisdictions across the country.

In Part I of this Essay, I offer an example of backdoor municipal immunity at work in Stewart v. City of Euclid, in which the Sixth Circuit granted qualified immunity to an officer who shot and killed an unthreatening, unresisting suspect because there was no prior court decision with sufficiently similar facts, then concluded that the grant of qualified immunity doomed the plaintiff’s failure-to-train claim against the City of Euclid, Ohio.14 In Part II, I describe a key finding from my recent study of use-of-force trainings in California and across the country that undercuts not only qualified immunity doctrine but also the backdoor municipal immunity allowed in Stewart: officers are trained about the broad outlines of their authority—through Supreme Court decisions like Graham v. Connor and Tennessee v. Garner, which set out general parameters for reasonable force under the Fourth Amendment15—but are not taught the facts and holdings of the kinds of cases that clearly establish the law for qualified immunity purposes.16 If officers are not trained about the facts and holdings of such cases, neither qualified immunity nor municipal liability should turn on whether a case with nearly identical facts exists. Part III argues that backdoor municipal immunity additionally subverts the deterrence and compensation rationales underlying the Supreme Court’s long-standing holding that local governments should not receive the protections of qualified immunity.

Backdoor municipal immunity is nonsensical and unjust. But it is not set in stone. Two circuits—the Ninth and Eleventh—have ruled that a Monell failure-to-train claim can proceed even if the individual officers received qualified immunity.17 Four circuits—the First, Fifth, Sixth, and Eighth—have adopted backdoor municipal immunity.18 Different panels of the Second and Tenth Circuits have ruled both ways.19 The issue has not squarely been addressed by the remaining circuits.20 In 2021, the Supreme Court declined to hear two cases in which municipalities received the protections of backdoor municipal immunity—Stewart and a case from the Tenth Circuit—which could have resolved this circuit split.21 It is anyone’s guess whether the Court will ever clarify this aspect of Monell. Until they do, lower courts are left to chart their own course. This Essay shows why courts should keep qualified immunity uncoupled from Monell.

I. backdoor municipal immunity in euclid, ohio and beyond

In early 2017, all of the officers employed by the Euclid, Ohio Police Department (EPD) attended a use-of-force training. In that training, the officers were shown a cartoon of an officer beating a prone civilian with the caption “protecting and serving the poop out of you.”22

Text  Description automatically generated with medium confidence

Officers were also shown a Chris Rock sketch called “How Not To Get Your Ass Kicked by the Police!” Among the “tips” Rock offered was: “If you have to give a friend a ride, get a white friend,” because “[a] white friend can be the difference between a ticket and a bullet in ya’.”23 Rock also joked that Rodney King “wouldn’t’ve got his ass kicked” if King had followed his advice.24

While the Department dedicated precious time in its training to this off-color humor, it appears to have spent little time instructing its officers about the constitutional limits on their authority to use force. The Department’s use-of-force training exercises were based on scenarios that never changed, and the officers’ performance in those exercises were never evaluated.25 At roll call, supervising officers would simply read the use-of-force policy to officers until “it [was] believed that all the officers ha[d] heard it.”26 The EPD offered no guidance in its policies or trainings about several common scenarios that its officers regularly confronted.27 One such area was how to remove people from cars—even though Euclid police officers routinely performed this task.28

A few months after the EPD’s tasteless and incomplete use-of-force training, in the early morning of March 13, 2017, Euclid police officer Matthew Rhodes shot and killed Luke Stewart, a twenty-three-year-old Black man, as Rhodes was trying to remove Stewart from Stewart’s car. Stewart was unarmed, unresisting, and unthreatening.29

The night before, Stewart had driven to a friend’s house, hoping to spend the night there. When the friend did not answer their phone, Stewart fell asleep. At 6:50 AM, Stewart’s friend’s neighbor called the police, reporting that there was a “creepy looking car . . . parked outside” of her house.30 Euclid police officer Louis Catalani came to the scene and saw Stewart asleep in his car. Officer Catalani checked the license plate and saw that the car’s owner had an outstanding warrant but could tell Stewart was too young to be the owner. Officer Catalani also saw a “digital scale” in the car and what appeared to be a marijuana blunt and the screw-on cap of a wine bottle.31 Officer Catalani then radioed Officer Rhodes, telling him that they were “goina [sic], uh, end up pulling this guy out.”32 A few minutes later, Officer Rhodes arrived on the scene and parked his car in front of Stewart’s.

Officer Catalani knocked on Stewart’s window, waved, and said “hi.”33 Stewart woke up, waved, sat up, and started his car—he had, apparently, thought the officer was communicating that Stewart should get going. Officer Catalani then yelled for Stewart to stop, opened the driver-side door, grabbed Stewart’s left arm, and tried to pull him out of the car. At the same time, Officer Rhodes opened the passenger-side door and began pushing Stewart. Stewart put his car in gear and started driving.

Stewart first hit Rhodes’s vehicle, then maneuvered around it. Next, Rhodes hopped into the passenger seat of Stewart’s moving car. Stewart looked over at Rhodes and asked, “Why are you in my car?”34 According to Officer Rhodes, Stewart did not threaten him, attempt to fight him, or speed up. He followed traffic laws, even stopping at an intersection.

Although Stewart posed no threat, Officer Rhodes yelled at Stewart, punched him in the head, and tased him six times during the first minute he was in Stewart’s passenger seat.35 Stewart drove his car onto a curb after Rhodes tased him. Then, when the car was stopped, Officer Rhodes shot Stewart twice in his torso. Only after being shot twice did Stewart offer any resistance, according to Rhodes, by attempting to “strike” Rhodes.36 Officer Rhodes then shot Stewart three more times, killing him. It had been less than two minutes from the time Officer Catalani first knocked on Stewart’s window and waved and Stewart waved back.

Mary Stewart, Luke Stewart’s mother, brought a Section 1983 suit against Rhodes and Catalani for using excessive force and against the City of Euclid for failing to train its officers properly. The district court found that the officers had not violated Stewart’s constitutional rights. But when Mary Stewart appealed the decision regarding Officer Rhodes, the Sixth Circuit disagreed. Although the court of appeals viewed the facts as showing “some danger to Rhodes,” it concluded that Officer Rhodes’s conduct was objectively unreasonable under the circumstances. Luke Stewart was not aggressive, had no weapon, and was not driving in a dangerous manner; the car had stopped and was in neutral when Rhodes fatally shot him.37

Although the Sixth Circuit found that Officer Rhodes had violated the Fourth Amendment, it nevertheless granted him qualified immunity after concluding that there was no “controlling authority or a robust consensus of cases of persuasive authority . . . [that] placed the constitutional question beyond debate.”38 As the Sixth Circuit recognized in its opinion, it is clearly “unreasonable to seize a fleeing felon with deadly force when the suspect poses no immediate threat to officers or others.”39 Indeed, as the panel noted, the Sixth Circuit had previously denied qualified immunity in circumstances very similar to those in Stewart, where officers shot at suspects driving away when there was no risk to the officers or others.40 But in those other cases, officers were standing outside the suspects’ cars when they fired their weapons; in no prior Sixth Circuit case had an officer taken the extraordinary step of getting inside the car of a fleeing suspect before shooting him.41 The singularity of Officer Rhodes’s choice paradoxically shielded him from liability. It meant that, according to the Sixth Circuit, no cases clearly established Stewart’s rights in the “particular circumstances” that Rhodes had faced, thus entitling him to qualified immunity.42

But the Sixth Circuit went still further, concluding that Officer Rhodes’s entitlement to qualified immunity also doomed Mary Stewart’s Monell failure-to-train claim against the City of Euclid. The Sixth Circuit criticized Euclid’s use-of-force trainings as “inappropriate and tasteless.”43 The court also observed that the trainings were lacking in rigor: “Even the components of the program that can be stomached appear skimped, such as the single genre of factual scenarios used to test officers.”44 Yet it held that a municipality’s failure to train its employees can only be a basis for liability if it reflects a “deliberate indifference to the rights of its inhabitants” and that such deliberate indifference is categorically impossible if that right has not been “clearly established” for qualified immunity purposes.45 In the view of the Sixth Circuit, because no prior case ruled that using deadly force while inside an unthreatening suspect’s vehicle violated the Fourth Amendment, the violation of Luke Stewart’s rights “cannot be the ‘known or obvious consequence’ disregarded by the City of Euclid through its training program.”46

Less than two months before the Stewart decision, the Sixth Circuit reached a contrary result in a remarkably similar case brought by the same team of lawyers against the City of Euclid.47 In that case, Wright v. City of Euclid, officers ordered Lamar Wright out of his car at gunpoint. Wright had a colostomy bag, so he moved slowly as he tried to comply with the officers’ demands; in response, the officers pepper-sprayed him at point-blank range and tased him in the abdomen.48 In Wright, the Sixth Circuit expressed disapproval of the very same use-of-force training materials that were submitted to the court in Stewart, describing them as “offensive” and concluding that “[a] reasonable jury could find that the City’s excessive-force training regimen and practices gave rise to a culture that encouraged, permitted, or acquiesced to the use of unconstitutional excessive force, and that, as a result, such force was used on Wright.”49

There was one critical difference between the cases that determined the divergent fates of Wright’s and Stewart’s Monell claims: the Sixth Circuit ruled that the officers in Wright’s case were not entitled to qualified immunity because other cases had made clear, to that panel’s satisfaction, that it is unconstitutional for an officer to point a gun at and tase a person who is not fleeing, posing a safety risk, or actively resisting arrest.50 Because the officers in Wright were denied qualified immunity, Wright could proceed with his failure-to-train claim against Euclid. Because the Sixth Circuit ruled Officer Rhodes was entitled to qualified immunity, Luke Stewart’s mother could not.

Courts around the country are divided about the relationship between qualified immunity and Monell failure-to-train claims. The First, Fifth, Sixth, and Eighth Circuits have embraced backdoor municipal immunity; the Ninth and Eleventh Circuits have rejected it; the Second and Tenth Circuits have vacillated on the issue; and the remaining circuits have not squarely addressed it.51 In the view of judges who favor backdoor municipal immunity, a grant of qualified immunity must foreclose a municipal liability failure-to-train claim because a city cannot be deliberately indifferent to the need for training about a right that is not clearly established.52 Judges taking the opposing view argue that evidence of obviously deficient training or a history of prior similar misconduct should be a sufficient basis for a failure-to-train claim, regardless of whether an officer in the case received qualified immunity53—and that backdoor municipal immunity amounts to a qualified immunity for local governments, a protection rejected by the Supreme Court in Owen v. City of Independence.54

Thus far, the debate over backdoor municipal immunity has not accounted for how officers are actually trained about the scope of their authority, or the impact of backdoor municipal immunity on the deterrence and compensation goals of Section 1983. The remainder of this Essay fills these gaps.

II. the fallacy of backdoor municipal immunity

Backdoor municipal immunity relies on a fundamental misunderstanding of the ways in which officers are trained about the scope of their constitutional authority—a misunderstanding that the Supreme Court’s qualified immunity decisions perpetuate.

The Supreme Court has repeatedly explained that plaintiffs must find a prior court decision with facts almost identical to those that the officer faced in the case at hand to defeat a qualified immunity motion.55 The Court has described this exacting standard as necessary to ensure that officers are on notice of the unconstitutionality of their conduct.56 It does not appear to be writing about constructive notice. Instead, the Court’s qualified immunity decisions seem to rest on two paired assumptions: first, that officers are actually educated about the facts and holdings of court opinions that apply general principles from cases like Graham and Garner, and second, that officers recall and contemplate those opinions when deciding whether and how to act on the job.57 As just one example of the Court’s reliance on these assumptions, the Court’s per curiam opinion in Kisela v. Hughes explained that factually similar precedent is necessary to clearly establish the law because “it is sometimes difficult for an officer to determine how the relevant legal doctrine . . . will apply to the factual situation the officer confronts” and “[p]recedent involving similar facts can help . . . provide an officer notice that a specific use of force is unlawful.”58

Backdoor municipal immunity makes some sense if one adopts the Supreme Court’s assumptions about how officers are notified of and rely on court decisions. If officers need factually similar precedent to put them on notice that a use of force is unlawful, and no factually similar precedent exists, then a city cannot be deliberately indifferent for failing to train them about their obligations under the circumstances they confronted.

But both the Supreme Court’s insistence that plaintiffs identify prior cases with virtually identical facts to defeat a qualified immunity motion and backdoor municipal immunity—which provides that a Monell failure-to-train claim cannot succeed if there is no prior case with virtually identical facts—bear scant resemblance to the ways in which police departments across the country actually train their officers. When I reviewed hundreds of California law enforcement use-of-force policy manuals, trainings, and other materials, I found that officers are regularly and repeatedly educated about Graham and Garner, which offer a general framework for understanding when force is appropriate, yet are virtually never educated about the facts and holdings of cases applying Graham and Garner to more specific fact patterns—the kinds of cases that clearly establish the law for qualified immunity purposes.59

More than three-fourths of the training outlines I reviewed reference no court decision applying Graham or Garner.60 Among the minority of training outlines that do reference cases applying Graham or Garner, the outlines include minimal detail about those cases’ facts and holdings. Instead, those cases are used to communicate broad principles that build on Graham and Garner—the notion, for example, that officers do not need to use the least available force so long as the force they do use is reasonable.61

Many manuals and trainings include a variety of scenarios intended to help officers understand the constitutional limits of their authority, but these scenarios are not drawn from court opinions. For example, the basic-training workbook for California officers describes the holdings of Graham and Garner, then describes various scenarios and what would constitute reasonable and unreasonable responses.62 The image below is one of the scenarios in the workbook—a traffic stop in which the officer discovers that the driver has several outstanding traffic warrants—and examples of reasonable and unreasonable actions the officer could take. The California basic-training workbook makes clear that the “reasonable” response described in the scenarios is one, but not the only, reasonable way to react.

Table  Description automatically generated

Far from preparing officers to respond to the multitude of scenarios they may face by training them about the facts and holdings of prior cases, the California basic-training workbook offers a few generic examples intended to help prepare officers “to select and use a response that is objectively reasonable under the totality of the facts and circumstances confronting the officer at the time.”63 The hundreds of California use-of-force training materials I reviewed took this same basic approach, and police-practices experts confirmed that California agencies’ training practices were consistent with those employed by departments nationwide.64

These findings reveal the folly of the Supreme Court’s insistence that qualified immunity motions can only be defeated by pointing to a case with nearly identical facts. Although the Court has confidently asserted that qualified immunity requires the existence of a prior factually similar case in order to “provide an officer notice that a specific use of force is unlawful,”65 law enforcement officers across the country do not actually learn the facts and holdings of the types of cases that clearly establish the law for qualified immunity purposes. Instead, officers are educated about general and well-established constitutional principles—such as the unconstitutionality of using force against a fleeing, but nonthreatening, suspect66—and then get comfortable applying these principles by considering a variety of factual scenarios unmoored from case law.

Evidence that police departments do not train their officers about the facts and holdings of cases that clearly establish the law for qualified immunity purposes also reveals the illogic of backdoor municipal immunity. In Stewart, the Sixth Circuit held that the plaintiff’s inability to point to a prior case in which an officer got into a car with a nonthreatening suspect and then shot him meant that the City of Euclid could not be held liable for failing to train its officers properly.67 But even if such a case had existed, neither Officer Rhodes nor the hundreds of thousands of other officers employed by departments around the country would have been taught its facts and holding. In Wright, the Sixth Circuit’s denial of qualified immunity to the officers who tased and pointed their gun at Wright meant that his failure-to-train claim against the City of Euclid could proceed.68 But Euclid officers and officers around the country almost certainly were not trained about the facts and holdings of the cases that clearly established Wright’s constitutional rights, either. The goal of police use-of-force trainings is for officers to develop the judgment to recognize when force is excessive, not to develop an encyclopedic knowledge of prior court decisions that they can recall and analogize to or distinguish from at a moment’s notice.

The Supreme Court has instructed that a Monell failure-to-train claim can be proven with evidence of a pattern of prior, similar constitutional violations known to final policymakers or if “the need for more or different training is so obvious . . . that the policymakers of the city can reasonably be said to have been deliberately indifferent to the need.”69 The Sixth Circuit concluded in Wright that Euclid’s use-of-force training was obviously deficient. As the court explained:

The Euclid Police Department’s training policy and procedures mandate that “[t]he department will establish and maintain a training committee.” However, no such training committee apparently has ever existed.

The City’s training seems to consist initially of simply reading the use-of-force policy to the officers at rollcall until “it is believed that all the officers have heard it,” which is then followed up with a one-or-two-page quiz that may or may not be given to officers. The City also engages in some sort of practical training exercise in which officers are given scenarios in which they may use force. But according to [the person] who implemented these scenario-based trainings, the scenarios never changed, and the officers’ performances were never evaluated. And recall that this training also included the graphic and comedy skit . . . .70

In addition, officers were never trained about how to respond to several common scenarios, including how to remove people from cars—the very situation that the officers in Stewart and Wright confronted.71 The Sixth Circuit in Wright concluded that “[a] reasonable jury could find that the City’s excessive-force training regimen and practice gave rise to a culture that encouraged, permitted, or acquiesced to the use of unconstitutional excessive force.”72 This evidence should have been sufficient for a failure-to-train claim to proceed against the City of Euclid in Stewart, as it was in Wright.

One could reach a very different conclusion from my study’s findings—that the problem is not with the Supreme Court’s precedent or with backdoor municipal immunity but with the ways in which police departments train their officers. One might believe that officers should be trained about the facts and holdings of cases that clearly establish the law for qualified immunity purposes. Yet even if one believes that departments should do more to incorporate court decisions into their trainings, no law enforcement agency could possibly educate their officers about all of the court decisions that have adjudicated the constitutionality of law enforcement officers’ conduct.73

Recall that the California basic-training workbook instructs officers about the holdings of Graham and Garner and then sets out a few scenarios, including one in which an officer has engaged a suspect at a traffic stop. If those responsible for designing California law enforcement officers’ basic training believed that officers were put on notice about the scope of their constitutional authority only by learning the facts and holdings of prior court decisions, the workbook’s treatment of reasonable and unreasonable uses of force during traffic stops might have looked like this:


subject’s action(s) and officer’s response(s)

court’s ruling

Donovan v. Phillips, 685 Fed. App’x 611 (9th Cir. 2017)

During a vehicle stop, Jennifer Donovan (the passenger) failed to comply with Officer Joshua Phillips’ instruction that she remain in the vehicle while Officer Phillips administered a field sobriety test to the driver. Officer Phillips placed Donovan in a control hold—he approached her, gripped her wrist, and pulled her arm downward, causing Donovan to roll to the ground.

Officer Phillips’ use of force was reasonable as a matter of law because he used minimal force and, under the circumstances, his interest in quickly securing her was significant.

Gahn v. Fujino, 39 F.3d 1187 (9th Cir. 1994)

Occupants of a vehicle fled from a police officer’s lawful stop. When Deputy Lauterbach approached the vehicle, Babcock refused to provide identification or exit the truck. Deputy Lauterbach used force to remove Babcock from the vehicle and arrest him. Babcock resisted Deputy Lauterbach’s arm hold. Officer Fujino then applied a chokehold, squeezing Babcock’s throat until he lost consciousness.

It was objectively reasonable for Deputy Lauterbach to use force sufficient to remove the resisting suspect from the vehicle in order to complete the arrest and control the situation. Based on the evidence at summary judgment, the chokehold applied by Officer Fujino was not objectively reasonable.

Coles v. Eagle, 704 F.3d 624 (9th Cir. 2012)

Harry Coles, a suspected car thief, whose car was boxed in between a concrete barrier and a patrol car, claimed he was unable to unlock his car door. Officers Joshua Eagle and Elton Robertson ordered Coles to exit the vehicle and keep his hands visible. Coles tried to exit the vehicle but could not, and then placed both hands on the wheel and looked straight ahead. Officer Eagle then smashed the driver’s side window with his baton, and the officers pulled Coles out of the car through the window. The officers kicked and beat Coles once they removed him from the car.

On these facts, it was objectively unreasonable for the officers to smash Coles’s car window and drag him through it, and to kick and beat Coles once they had removed him from the car.

The facts and holdings of these three cases barely scratch the surface of the cases officers would need to learn. By my count, as of July 10, 2020, there were 284 Supreme Court and Ninth Circuit decisions that interpret Graham and Garner and that could be used to clearly establish the law for qualified immunity purposes.74 If police trainers only spent five minutes explaining the facts and holdings of each of these cases, it would take almost the entirety of the time allotted for officers’ biannual trainings in California.75 Officers would also need to learn the facts and holdings of hundreds or thousands of additional cases—those from their home circuit concerning officers’ constitutional authority in other contexts, as well as cases from other circuits.76 Although a 2022 study from the Institute of Justice indicates that the Ninth Circuit has issued the most decisions clearly establishing the law, Ohio is in the Sixth Circuit, which came in a close second—and, regardless, learning all relevant cases in sufficient detail would not be feasible in any circuit.77

Moreover, even if departments dedicated sufficient time to teaching officers about all of these cases, there is no way that a law enforcement officer could remember detailed information about their facts and holdings. And, even if they somehow could, all that we know about decision-making in high-speed, high-stress circumstances suggests they would not be able to recall those facts and holdings in the moments that lead up to a decision about whether or not to use force.78

The Supreme Court is usually very willing to defer to law enforcement agencies’ assertions of expertise79—including about the ways in which they train their officers.80 It follows that the Court should defer to the general consensus among law enforcement trainers and policymakers that officers do not need to be educated about the facts and holdings of all of the cases applying Graham and Garner to understand the scope of their constitutional authority to use force. Section 1983 doctrine should adjust to reflect this consensus as well. If departments limit the scope of officers’ use-of-force policies and trainings to broad constitutional principles like those articulated in cases like Graham and Garner, courts should define qualified immunity’s “clearly established law” standard at that higher level of generality. Similarly, the fate of a Monell claim alleging failure to train should not turn on the existence of a prior court decision with virtually identical facts. Instead, it should turn on the adequacy of a city’s training for its officers regarding the general constitutional principles at issue in the case.

If qualified immunity were defined in this broader way and granted to officers only when the law was truly unsettled, it might make sense for findings of qualified immunity to prompt conclusions that policymakers were not deliberately indifferent for failing to train their officers about that unsettled area of the law. But so long as officers are granted qualified immunity unless plaintiffs can point to a case with nearly identical facts, the doctrine bears no relationship to the ways in which police departments actually train their officers, and a grant of qualified immunity should have no impact on a court’s analysis of a Monell failure-to-train claim.

III. the injustice of backdoor municipal immunity

Courts should also reject backdoor municipal immunity for another reason: it undermines the key rationales behind the Supreme Court’s long-standing instruction in Owen v. City of Independence that local governments should not receive the protections of qualified immunity.81

Tying Monell liability to qualified immunity mutes the deterrent power of Monell claims. The Owen Court believed that actions for damages against municipalities—without the protections of qualified immunity—could “serve as a deterrent against future constitutional deprivations”:

The knowledge that a municipality will be liable for all of its injurious conduct, whether committed in good faith or not, should create an incentive for officials who may harbor doubts about the lawfulness of their intended actions to err on the side of protecting citizens’ constitutional rights. Furthermore, the threat that damages might be levied against the city may encourage those in a policymaking position to institute internal rules and programs designed to minimize the likelihood of unintentional infringements on constitutional rights.82

The Owen Court expected that the threat of municipal liability would encourage jurisdictions like the City of Euclid to design rules and programs that made constitutional violations less likely to occur. But if a grant of qualified immunity forecloses a Monell failure-to-train claim, departments can administer racist and tasteless trainings, neglect to train officers about basic elements of their authority, rack up long records of prior misconduct incidents, and escape any responsibility for their malfeasance so long as the city’s policymakers are fortunate enough to have their officers violate people’s rights in previously unheard-of ways.

Backdoor municipal immunity also undermines the compensatory goals of Section 1983. The Owen Court explained that municipalities should not enjoy qualified immunity in part because “a damages remedy against the offending party is a vital component of any scheme for vindicating cherished constitutional guarantees,” and if both officers and governments were entitled to qualified immunity, “many victims of municipal malfeasance would be left remediless.”83

Backdoor municipal immunity delivers this same unjust fate. As Judge Donald explained in her partial concurrence and dissent in Stewart, “Luke Stewart should be alive today. He was unarmed, unsuspected of committing a serious felony, and behind the wheel of a stationary vehicle when Rhodes opened fire into his torso, chest, neck, and wrist.”84 Judge Donald was of the belief that “[q]ualified immunity should not shield Rhodes from the consequences of that unreasonable decision.”85 I agree. But if qualified immunity is going to shield Rhodes from the consequences of his actions, it is doubly unjust for the doctrine to further shield the City of Euclid from any responsibility under Section 1983 to remedy Luke Stewart’s death.

Although backdoor municipal immunity is unquestionably inconsistent with Monell’s deterrence and compensation goals, its effects will differ from case to case. As I have previously observed, the outcome of a civil rights plaintiff’s case cannot be pinned to a single doctrine—instead, it is the product of the legal ecosystem in which the case is brought, made up of federal and state laws, state and local actors, and information practices (like litigation, settlement, and indemnification decisions) that vary by jurisdiction.86 Backdoor municipal immunity is but one of many doctrines that can make it difficult for plaintiffs to succeed, and its impact will depend in part on other aspects of the civil rights ecosystem.

For example, backdoor municipal immunity has not foreclosed efforts to seek justice against the City of Euclid, Ohio. Although the grant of qualified immunity to Officer Rhodes prevented Mary Stewart from litigating her Section 1983 claims against the City of Euclid, the City has not escaped responsibility for its tasteless and incomplete trainings. Lamar Wright pursued his Monell failure-to-train claim against the City of Euclid and ultimately settled his Section 1983 claims against the involved officers and the City for $475,000.87 As his lawyers explained in the announcement of the settlement, Wright’s suit “led to EPD police training materials appearing in statewide and national media” and exposed the City’s unlawful training and failure to discipline its officers.88 Lamar Wright offered this comment about the settlement:

I filed this case to stand up against police brutality, and to stand with other victims of senseless attacks by officers from the Euclid Police Department. These officers’ illegal treatment of people in the city must stop. We need justice for all the victims of the EPD, and I hope that my case will lead to justice and change.89

Backdoor municipal immunity also did not completely foreclose Mary Stewart’s efforts to seek justice. Although she was prevented from seeking compensation from the City of Euclid and Officer Rhodes under Section 1983, her complaint also alleged various state-law claims against Officer Rhodes, including wrongful death, intentional infliction of emotional distress, and assault and battery. Stewart’s lawyers are continuing to litigate those claims in state court and are scheduled to go to trial in October 2022.90

But backdoor municipal immunity has nonetheless made justice more difficult for Mary Stewart to attain. Although the City of Euclid will likely pay any settlement or judgment against Officer Rhodes, Stewart cannot sue the City directly under Ohio state law.91 And while her attorneys could have recovered their reasonable fees had they prevailed after trial on the Section 1983 claim, Ohio law does not allow for fee shifting—so even if Stewart wins in state court, her attorneys will need to be paid a portion of her award.92 For these reasons, even if Mary Stewart prevails on her state-law claims, backdoor municipal immunity will have partially shielded the City of Euclid from responsibility and reduced the remedy she receives.

In other places, and other cases, backdoor municipal liability can have an even more devastating impact. Consider, for example, the civil rights case brought by Gloria Bustillos, a U.S. citizen who, on the afternoon of September 19, 2013, traveled from El Paso, Texas, to Juarez, Mexico to deliver food and clothes to her long-time friend.93 After delivering the food and clothes and visiting with her friend and her friend’s family, fifty-six-year-old Bustillos returned to the United States via the Paso del Norte Bridge in El Paso at approximately 6:30 PM.94 The Fifth Circuit describes what happened next:

After presenting her passport to Customs and Border Protection agents, Bustillos was immediately taken into custody despite telling agents that she was not in possession of narcotics. An increasingly intrusive series of searches followed.

First, two female agents conducted a pat down. The agents found no drugs. The agents then held Bustillos for a K-9 search. The K-9 failed to alert to the presence of drugs. Two agents then took Bustillos into a restroom, where they ordered her to pull down her pants and underwear and bend over slightly. The agents conducted a visual inspection of Bustillos’ vaginal and anal area. Again, the agents found no drugs. Despite no evidence of drugs, the agents placed tape on Bustillos’ legs and abdomen, handcuffed her, and transported her to the University Medical Center (the “Hospital”) in El Paso.

At the Hospital, Doctors Michael Parsa and Daniel Solomin (the “Doctors”) ordered a series of x-rays to search for drugs. The x-rays revealed no drugs. The Doctors then performed a pelvic exam. Again, the pelvic exam evidenced no drugs. Solomin then conducted a rectal exam. Yet again, Solomin found no evidence of drugs. As part of these searches, the Doctors, and Nurses Lynette Telles and Frank Mendez (the “Nurses”), allegedly “brutally” probed Bustillos’ cavities in the presence of hospital personnel. Bustillos did not consent to any of the above searches.

At approximately 4:00 a.m. the next morning, after finding no evidence of narcotics, the Doctors released Bustillos to CBP agents, who drove Bustillos to the international bridge and released her.95

Bustillos sued the federal Border Patrol agents and the doctors and nurses at the hospital for violating her constitutional rights, and she brought a Monell claim against El Paso County Hospital District/University Medical Center (UMC) for failing to train its staff properly. Bustillos’s complaint alleged that UMC had a policy prohibiting searches by hospital personnel unless they had a “valid, written, and signed informed consent form or a search warrant.”96 And although the district court denied requests by Bustillos for discovery to support her Monell failure-to-train claim against UMC, Bustillos introduced evidence from a case brought by another woman who had been subjected to body-cavity searches by one of the same doctors—Parsa—less than a year prior.97 In that prior case, Parsa testified that UMC offered no trainings or guidance of any kind to doctors or nurses about how to respond if law enforcement asks them to perform examinations.98 Parsa also testified that he was unaware of the hospital policy prohibiting examinations without a warrant or consent.99

But as far as the Fifth Circuit was concerned, neither the lack of training offered to UMC doctors and nurses about how to respond to warrantless search requests by law enforcement officers, nor the allegedly unlawful search conducted by one of the same medical providers less than a year prior, was relevant to Bustillos’s failure-to-train claim against UMC. The district court granted the doctors and nurses qualified immunity, and the Fifth Circuit affirmed, reasoning that “Bustillos’ allegations could potentially assert a constitutional violation” but that the officers were entitled to qualified immunity because Bustillos had not “carried her burden of pointing this panel to any case that shows, in light of the specific context of this case, that the Doctors’ or Nurses’ conduct violated clearly established law.”100 And because the Fifth Circuit has adopted backdoor municipal immunity, the grant of qualified immunity to the doctors and nurses doomed Bustillos’s failure-to-train claim.101

Backdoor municipal immunity was profoundly consequential in Gloria Bustillos’ case: although Mary Stewart could sue the City of Euclid under state law, Bustillos had no state-law claims to pursue. Texas state tort claims against local governments and government officials are extremely limited,102 and the state-law claims Bustillos included in her complaint were dismissed by the district court.103 Bustillos’s remaining constitutional claims against the federal and local-government officials were dismissed as well.104 Bustillos received nothing despite being held for ten hours and invasively searched without cause or consent. Even if there were no backdoor municipal liability in the Fifth Circuit, her Monell claim may not have succeeded—when I interviewed Texas civil rights lawyers, several reported that Monell claims are almost always unsuccessful.105 But backdoor municipal immunity nevertheless foreclosed what may have been Bustillos’s best chance for relief.

When backdoor municipal immunity makes justice harder—or impossible—to obtain, it dampens local governments’ incentives to properly train their officers and denies full compensation for people whose rights have been violated. Adding insult to injury, backdoor municipal immunity undermines these compensatory and deterrence goals of Section 1983—and the rationale in Owen—based on a clear misunderstanding of the ways local governments educate officers about the scope of their power.


As the Sixth Circuit found in Wright, a reasonable jury could conclude that the Euclid Police Department’s tasteless and crass use-of-force training exhibited deliberate indifference to the rights of its citizens and led to the excessive force used against Lamar Wright as he was pulled from his car. The same should have been true in Stewart. Officer Rhodes should not have received qualified immunity because the Supreme Court and the Sixth Circuit have both made clear that it is unconstitutional to use fatal force against an unresisting suspect. But regardless of whether Officer Rhodes was granted qualified immunity, qualified immunity should not have foreclosed Stewart’s Monell claim. Police departments do not train their officers about the facts and holdings of the kinds of cases that clearly establish the law for qualified immunity purposes, so courts’ assessment of failure-to-train claims should not turn on whether such cases exist. When the Sixth Circuit dismissed Stewart’s Monell claim, it compounded the illogic and injustice of the qualified immunity defense.

Thankfully, backdoor municipal immunity is not settled law. In this Conclusion, I suggest a few possible ways to correct course.

In 2021, the Supreme Court declined to grant certiorari in both Stewart and a Tenth Circuit case that had dismissed Monell failure-to-train claims because the individual officers were granted qualified immunity.106 Unless and until the Court steps in, lower courts remain free to develop this area of Section 1983 law. This Essay demonstrates the sensibility of those circuits that have rejected backdoor municipal liability; courts yet to confront the issue should follow their lead, and courts that have thus far embraced backdoor municipal immunity should reconsider their position moving forward.

Federal, state, and local legislators could also enact laws that would do away with backdoor municipal immunity. In 2020 and 2021, Congress considered, but ultimately failed to pass, the George Floyd Justice in Policing Act; the Act’s elimination of qualified immunity was a key sticking point in negotiations.107 But Senator Tim Scott, who led the Republicans’ negotiations, was amenable to the idea that local governments should be held vicariously liable for the unconstitutional conduct of their officers.108 If congressional appetite for police reform returns, there could be less resistance to a bill that kept qualified immunity intact but also clarified that a grant of qualified immunity should not absolve municipalities of responsibility for properly training their officers. Given the injustice of qualified immunity doctrine, this would only be a partial victory; but it could make a meaningful difference in cases like Stewart and Bustillos.

State and local legislatures can also take steps that would end backdoor municipal immunity, and some already have. In the wake of George Floyd’s murder in May 2020, state legislatures around the country considered bills to create state-law causes of action for constitutional violations by law enforcement officers and other government officials without the protections of qualified immunity.109 Where this type of legislation succeeds—including Colorado and New York City, thus far110—there is no qualified immunity for state-law claims against government officers and so, by necessity, no backdoor municipal immunity. Legislatures wary of eliminating qualified immunity for their officers could choose, instead, to make local governments vicariously liable for the conduct of their officers and clarify that the municipality is not entitled to the protections of qualified immunity. This type of legislation passed in New Mexico and effectively does away with both qualified immunity and backdoor municipal immunity in that state.111

Bills like Colorado’s have been strenuously opposed by law enforcement and union representatives who threaten that, without qualified immunity, officers will be bankrupted for split-second mistakes.112 As a result, these types of bills have failed in state legislatures far more often than they have succeeded.113 New Mexico’s approach, making local governments vicariously liable for constitutional violations by their officers, may be more politically palatable to law enforcement and union officials because individual officers can be assured they will not bear financial responsibility for settlements and judgments against them. On the other hand, New Mexico’s approach may be opposed by civil rights advocacy groups or victims of misconduct who believe that officers should be required to pay when they violate the Constitution. Government officials might also argue in opposition to bills like New Mexico’s on the grounds that they will increase the financial burden on cities, counties, and towns.

Time will tell whether bills adopting Colorado’s or New Mexico’s approaches will continue to be introduced in statehouses and whether they can garner enough support to get passed and signed into law. Time will also tell whether the Supreme Court will clarify the relationship between qualified immunity and municipal liability, and how lower courts will navigate this unsettled area of the law in the meantime. Given the lack of legislative success, and the Supreme Court’s decision not to grant certiorari in Stewart in 2021, I place my hope primarily in the lower courts. But while I am equivocal about whether backdoor municipal immunity will ultimately be rejected, I am absolutely certain that it should be. As this Essay shows, backdoor municipal immunity rests on a misunderstanding of how officers are trained about the scope of their authority, undermines well-settled Supreme Court precedent holding that municipalities are not entitled to qualified immunity, and subverts the deterrence and compensation goals of Section 1983. Backdoor municipal immunity must go, no matter how it gets gone.

Professor of Law, UCLA School of Law. Many thanks to Terry Gilbert and Jacqueline Greene, attorneys at Friedman, Gilbert + Gerhardstein, for providing me with additional details about the Stewart and Wright cases, and to Alex Reinert for thoughtful comments about a prior draft. Thanks also to the editors of the Yale Law Journal Forum for terrific editorial assistance, including Alan Chen, Nathan Cummings, Milo Hudson, Cynthia Long, and Angela Uribe.