The City of San Antonio appeals the district court’s entry of judgment on a jury verdict holding it responsible for an illegal arrest. Concluding that there was insufficient evidence to support municipal liability, we reverse and render judgment in the City’s favor.

I. FACTUAL AND PROCEDURAL BACKGROUND

This suit arises from the arrest of Robert Milam at the hands of employees of the City of San Antonio. While walking across a golf course on an evening in December 1999, Milam was stopped and detained by a San Antonio park ranger, Officer Escobedo. Park rangers are licensed peace officers, though with less training and authority than police officers. Escobedo says that his knowledge of past mischief committed on the golf course after dark led him to suspect that something was amiss, but Milam says that the ranger did not tell him why he was being detained. Milam and Escobedo each accuse the other of using abusive language during the encounter. Soon a second park ranger, Officer Coronado, arrived on the scene. According to Milam, the two park rangers roughly handcuffed him, then put him in one of the patrol cars. Eventually, Milam was turned over to Officer Land of the San Antonio Police Department 1 and taken to the police station. Milam was brought before a magistrate and charged with resisting arrest, though he was not charged with any underlying offense that had required the arrest in the first place. He spent the night in the county jail. When Milam later appeared for his arraignment, he found that the district attorney had dropped the charges for insufficient evidence.

Milam was quite upset by his treatment and wrote letters to many City officials telling them of his experience and urging them to take action to fix the apparent problems with their rangers and police. In response, the City’s Municipal Integrity Division began an investigation into the park rangers’ conduct, and the Police Department’s Internal Affairs Division started an inquiry into Land’s conduct. The Police Department later informed Milam that it had completed its investigation and had concluded that Land did not violate any rules. Milam did not hear back from City officials regarding the results of the City’s separate investigation of the park rangers; the Municipal Integrity Division stopped this investigation, on the advice of its risk-management staff, when it received notice that Milam was planning to file suit. Milam did in fact sue the City in the district court, asserting a § 1988 cause of action as well as state-law claims.

After dismissal of the state-law claims on immunity grounds, the § 1983 claim proceeded to trial. Milam produced evidence tending to show that his detention and arrest were unjustified. Milam sought to hold the City liable for its employees’ illegal conduct by introducing evidence that City policymakers were aware of and were indifferent to a pattern of illegal arrests by park rangers, that the rangers were inadequately trained and supervised, and that the City failed to respond meaningfully to Milam’s complaints. The City moved for judgment as a matter of law at the close of Milam’s case and again at the *624 close of the evidence, but the court denied the motions and sent the ease to the jury.

The jury found that the arrest was illegal, and the City does not challenge that finding. For purposes of the present appeal, two of the questions on the verdict form—both relating to municipal liability for the illegal arrest—are relevant. In Question 2, the jury was asked the following:

Do you find from a preponderance of the evidence that the city of San Antonio was consciously and deliberately indifferent to intentional and illegal arrests of individuals without probable cause by its park rangers, condoning a pattern or practice of such arrests by its park rangers?

In Question 3, the jury was asked the following:

Do you find from a preponderance of the evidence that the City’s policy-making authority, ratified the wrongful conduct of its officers in violation of Mr. Milam’s constitutional rights? 2

The jury answered “no” to Question 2 and “yes” to Question 3. Pursuant to the verdict form’s directive that the jury should proceed to consider damages if it answered “yes” to either Question 2 or Question 3, the jury awarded $100,000.

After the trial, the City again moved for judgment as a matter of law. The district court again denied the motion. The court later awarded Milam attorneys’ fees based on his status as a prevailing party. The City now appeals.

II. ANALYSIS

The City argues that there is no legally sufficient basis for the jury’s affirmative answer to Question 3, which presented a ratification theory of municipal liability. It therefore asks that we reverse and render judgment in its favor. Alternatively, it requests a new trial on the ground that the district court erroneously instructed the jury that city managers and department heads, not just the City Council, were policymakers who could expose the City itself to liability. We find that we need not address the City’s alternative request for a new trial, for we agree with its argument that there was insufficient evidence to support the ratification verdict.

A. Standard of Review

We review the district court’s ruling on the City’s motion for judgment as a matter of law de novo, applying the same Rule 50 standard as did the district court. See Coffel v. Stryker Corp., 284 F.3d 625, 630 (5th Cir. 2002). Judgment as a matter of law is appropriate with respect to an issue if “there is no legally sufficient evidentiary basis for a reasonable jury to find for [a] party on that issue.” Fed.R.Civ.P. *625 50(a)(1). This occurs when the facts and inferences point so strongly and overwhelmingly in the movant’s favor that reasonable jurors could not reach a contrary verdict. Coffel, 284 F.3d at 630. In considering a Rule 50 motion, the court must review all of the evidence in the record, drawing all reasonable inferences in favor of the nonmoving party; the court may not make credibility determinations or weigh the evidence, as those are jury functions. Reeves v. Sanderson Plumbing Prods., Inc., 530 U.S. 133, 150, 120 S.Ct. 2097, 147 L.Ed.2d 105 (2000). In reviewing the record as a whole, the court “must disregard all evidence favorable to the moving party that the jury is not required to believe. That is, the court should give credence to the evidence favoring the nonmovant as well as that evidence supporting the moving party that is uncontradicted and unimpeached, at least to the extent that that evidence comes from disinterested witnesses.” Id. at 151, 120 S.Ct. 2097 (citation and internal quotation marks omitted).

B. Principles of Municipal Liability

Milam’s suit names no employees of San Antonio but only the municipality itself. The Supreme Court held in Monell v. Department of Social Services that municipalities are “persons” subject to suit under 42 U.S.C. § 1983. 436 U.S. 658, 663, 690, 98 S.Ct. 2018, 56 L.Ed.2d 611 (1978). At the same time, however, the Court ruled that municipalities are not liable on a respondeat superior basis; that is, a municipality cannot be held liable simply by virtue of the fact that one of its employees violated a person’s federal rights. Id. at 691, 98 S.Ct. 2018. For a municipality to be liable, the municipality itself must cause the violation through its policies. “[I]t is when execution of a government’s policy or custom, whether made by its lawmakers or by those whose edicts or acts may fairly be said to represent official policy, inflicts the injury that the government as an entity is responsible under § 1983.” Id. at 694, 98 S.Ct. 2018.

Municipal policy can take several guises. Monell itself involved an acknowledged, formal policy. See id. at 690, 98 S.Ct. 2018. But to say that municipal policy itself must cause the injury is not to say that policy is limited to formal pronouncements, ordinances, and the like. The Court’s opinion in Monell recognized, in keeping with the language of § 1983, that a plaintiff can sue a municipality for deprivations “visited pursuant to governmental ‘custom’ even though such a custom has not received formal approval through the body’s official decisionmaking channels.” Id. at 691, 98 S.Ct. 2018. Cases have recognized that municipal custom can sometimes be proven through evidence of a persistent pattern of conduct. See City of St. Louis v. Praprotnik, 485 U.S. 112, 127, 108 S.Ct. 915, 99 L.Ed.2d 107 (1988) (plurality opinion) (referring to “a widespread practice that, although not authorized by written law or express municipal policy, is so permanent and well settled as to constitute a ‘custom or usage’ with the force of law” (internal quotation marks omitted)); Bennett v. City of Slidell, 728 F.2d 762, 767-68 (5th Cir. 1984) (en banc). That is, the existence of a persistent pattern of illegal conduct, tolerated by municipal policymakers, 3 tends to show that the subject conduct does not represent an unauthorized departure from lawful policy but instead represents the realization of an unlawful policy. Thus, we have on several occasions upheld findings of municipal lia *626 bility that were predicated on patterns of illegality that rose to the level of customary policy. See, e.g., Lawson v. Dallas County, 286 F.3d 257, 263-64 (5th Cir. 2002). But cf. Pineda, 291 F.3d at 329 (finding insufficient evidence of a municipal custom of illegal conduct).

Municipalities can also be liable, in certain situations, for single episodes of conduct that are not part of any pattern of illegality. See generally Bd. of County Comm’rs v. Brown, 520 U.S. 397, 405-06, 117 S.Ct. 1382, 137 L.Ed.2d 626 (1997) (summarizing the Court’s single-episode cases). For example, plaintiffs can hold municipalities liable for single instances of conduct perpetrated by the policymakers themselves; such one-time conduct can represent official “policy” even though it does not necessarily form part of a plan or rule developed to govern all like occasions. See, e.g., Pembaur v. City of Cincinnati, 475 U.S. 469, 480-81, 484-85, 106 S.Ct. 1292, 89 L.Ed.2d 452 (1986) (holding a county liable where the county prosecutor, a policymaker, directed deputy sheriffs to forcibly serve a capias in violation of the Fourth Amendment). In addition, in Praprotnik the Supreme Court described a scenario in which a municipality could be held liable for a single episode of conduct initiated by a non-policymaker employee. The plaintiff in Prapj’otnik claimed that his supervisors in the city government had retaliated against him, and, through the supervisors’ actions, he hoped to hold the city itself liable. Looking to state law, the plurality opinion determined that the municipality’s policymaking authority over employment was vested in the mayor, the aldermen, and the civil service commission, but not in the plaintiffs’ supervisors. 485 U.S. at 124-26, 128-29, 108 S.Ct. 915. It therefore rejected the plaintiffs claim, but in the course of doing so it noted certain factors that could have changed the outcome:

It would be a different matter if a particular decision by a subordinate was cast in the form of a policy statement and expressly approved by the supervising policymaker.... In [such a case], the supervisor could realistically be deemed to have adopted a policy that happened to have been formulated or initiated by a lower ranking official.

Id. at 130, 108 S.Ct. 915. Elsewhere, the opinion included the following comments about how a policymaker could “ratify” a subordinate’s illegal conduct, thus putting the force of municipal policy behind it:

[Wjhen a subordinate’s decision is subject to review by the municipality’s authorized policymakers, they have retained the authority to measure the official’s conduct for conformance with their policies. If the authorized policymakers approve a subordinate’s decision and the basis for it, their ratification would be chargeable to the municipality because their decision is final.

Id. at 127,108 S.Ct. 915.

This sort of ratification is most readily conceptualized in contexts like employment. For example, if a school board—a policymaker under Monell—approves a superintendent’s decision to transfer an outspoken teacher, knowing of the superintendent’s retaliatory motive for doing so, the governmental entity itself may be liable; but if the school board lacks such awareness of the basis for the decision, it has not ratified the illegality and so the district itself is not liable. See Beattie v. Madison County Sch. Dist., 254 F.3d 595, 603-05 (5th Cir. 2001) (discussing and distinguishing Harris v. Victoria Independent School District, 168 F.3d 216, 225 (5th Cir. 1999)).

It is important to recognize that the ratification theory, in whatever context it arises, is necessarily cabined in several *627 ways. Praprotnik itself recognized that policymakers who “[sjimply go[ ] along with” a subordinate’s decision do not thereby vest final policymaking authority in the subordinate, nor does a “mere failure to investigate the basis of a subordinate’s discretionary decisions” amount to such a delegation. 485 U.S. at 130, 108 S.Ct. 915. Such limitations on municipal liability are necessary to prevent the ratification theory from becoming a theory of respondeat superior, which theory Monell does not countenance. See id. at 126, 108 S.Ct. 915 (“If the mere exercise of discretion by an employee could give rise to a constitutional violation, the result would be indistinguishable from respondeat superior liability.”); cf. City of Oklahoma City v. Tuttle, 471 U.S. 808, 830-31, 105 S.Ct. 2427, 85 L.Ed.2d 791 (1985) (Brennan, J., concurring in the judgment). Policymakers alone can create municipal liability, and so any violation must be causally traceable to them, not just to their subordinates.

C. Application to Milam’s Case

The evidence adduced at trial might have provided a legally sufficient basis for the jury to determine that the City’s policymakers had tolerated a pattern of illegal arrests that rose to the level of customary policy. The jury, though, specifically rejected a pattern-and-practice theory in its negative answer to Question 2. Milam is therefore left with the task of trying to hang the evidence presented at trial onto the doctrinal hooks of the ratification theory. It is not an easy fit because, at least facially, an illegal arrest that is completed without the involvement of any policymaker does not look like the typical situation in which a policymaker could “approve! ] [the employee’s] decision and the basis for it” such that municipal policy can be said to have caused the harm. Praprotnik, 485 U.S. at 127, 108 S.Ct. 915. Milam attempts in a few different ways to provide evidentiary support for the ratification verdict, but we conclude that the evidence does not support this theory of liability.

Milam’s primary argument is that his ratification theory is aimed at situations in which policymakers have tacitly permitted informal practices to rise to the level of official municipal policy. 4 It is certainly true, as we discussed above, that Monell recognizes that informal customs and usages, no less than formally promulgated pronouncements and ordinances, can come to represent a type of municipal policy. See Monell, 436 U.S. at 690-91, 98 S.Ct. 2018. Actions taken pursuant to such a customary policy can then subject the municipality to § 1983 liability. Nonetheless, this does not help Milam’s case. If Question 2 on the verdict form had limited the jury to considering whether the City had a policy of the formal-pronouncement type, then perhaps Milam’s evidence that the City had allowed a pattern of illegal arrests could be shoe-horned into Question 3, the ratification interrogatory. But Question 2 was not so limited; rather, it fully contemplated the possibility that the City had tacitly adopted a customary policy. It did not ask the jury whether the City had promulgated ordinances or the like, but it instead asked them whether the City had “condon[ed] a pattern and practice” of illegal arrests. The jury answered that it had not. Milam’s attempt to equate ratification with liability for customary policy strips the ratification theory of any independent content within the circumstances of this case.

*628 Pursuing another tack, Milam also points to evidence concerning how the City-responded to his arrest as support for his argument that the City “ratified” the arresting officers’ conduct. Milam sent letters to many of San Antonio’s officials, including some who were policymakers; no disciplinary action against the rangers resulted. We do not think that this permits an affirmative answer to Question 3. First, this record does not present a situation where the policymakers have approved the “decision and the basis for it.” Praprotnik, 485 U.S. at 127, 108 S.Ct. 915; see also id. at 130, 108 S.Ct. 915 (referring to a situation in which “a particular decision by a subordinate was cast in the form of a policy statement and expressly approved by the supervising policymaker”). That the policymakers failed to take disciplinary action in response to Milam’s complaints does not show that they knew of and approved the illegal character of the arrest, determining that it accorded with municipal policy. See id. at 130, 108 S.Ct. 915. Second, it is hard to see how a policymaker’s ineffectual or nonexistent response to an incident, which occurs well after the fact of the constitutional deprivation, could have caused the deprivation. See Thomas ex rel. Thomas v. Roberts, 261 F.3d 1160, 1174-75 (11th Cir. 2001), vacated, 536 U.S. 953, 122 S.Ct. 2653, 153 L.Ed.2d 829 (2002), reinstated, 323 F.3d 950 (11th Cir. 2003); Vukadinovich v. McCarthy, 901 F.2d 1439, 1444 (7th Cir. 1990).

To be clear, we do not say that ladduster disciplinary responses are never relevant in a Monell case and can never cause constitutional injuries. First, municipal policymakers who fail to supervise and to discipline their police officers, acting with deliberate indifference to the citizens’ rights, could create municipal liability if the lack of supervision then caused a deprivation. Cf. Brown, 520 U.S. at 406-10, 117 S.Ct. 1382 (discussing liability for inadequate training and hiring policies). Second, even though a policymaker’s response to a particular incident may not cause the injury, the response might provide evidence of the content of a municipality’s policies. That is, the failure to take disciplinary action in response to an illegal arrest, when combined with other evidence, could tend to support an inference that there was a preexisting de facto policy of making illegal arrests: the policymaker did not discipline the employee because, in the policymakers’ eyes, the employee’s illegal conduct actually conformed with municipal policy. See Bordanaro v. McLeod, 871 F.2d 1151, 1166-67 (1st Cir. 1989); Grandstaff v. City of Borger, 767 F.2d 161, 171 (5th Cir. 1985); 5 cf. Praprotnik, 485 U.S. at 131, 108 S.Ct. 915 (“Refusals to carry out stated policies could obviously help to show that a municipality’s actual policies were different from the ones that had been announced.”). But, once again, these are all possibilities that might have supported an affirmative answer to Question 2, which the jury declined to give. The unsatisfactory investigation was not an actionable “ratification,” if that theory is to have any independent content.

Milam also presented evidence that the City failed to train the park rangers properly, and the district court instructed the jury on a failure-to-train theory. See generally City of Canton v. Harris, 489 U.S. 378, 109 S.Ct. 1197, 103 L.Ed.2d 412 (1989) (recognizing and describing this theory of *629 liability). The verdict form, however, did not include a question that contained failure-to-train language. If either question was able to embrace the theory, it would be Question 2: The district court’s failure-to-train instruction in part tracked the language in Question 2, 6 and failure-to-train cases typically involve policymakers who display indifference to a pattern of tortious conduct that makes the need for training obvious. See Brown, 520 U.S. at 407-10, 117 S.Ct. 1382. In any case, the failure-to-train theory is not the ratification theory discussed in Praprotnik. Once more, we must conclude that there is insufficient evidence to support the jury’s answer to Question 3.

D. Attorneys’ Fees Award

Milam was awarded attorneys’ fees under 42 U.S.C. § 1988(b) based on his status as a prevailing party in this civil-rights case. Given our disposition of the case, he is no longer a prevailing party and so the award cannot stand. See Johnson v. Rodriguez, 110 F.3d 299, 316 (5th Cir. 1997).

III. CONCLUSION

For the foregoing reasons, we REVERSE the district court’s judgment in Milam’s favor, VACATE the award of attorneys’ fees, and RENDER judgment in the City’s favor that Milam take nothing on his complaint.