(delivered the opinion of the Court)
This is a Bivens action against criminal investigators for inducing prosecution in retaliation for speech. The question is whether the complaint states an actionable violation of the First Amendment without alleging an absence of probable cause to support the underlying criminal charge. We hold that want of probable cause must be alleged and proven.
In the 1980’s, respondent William G. Moore, Jr., was the chief executive of Recognition Equipment Inc. (REI), which manufactured a multiline optical character reader for interpreting multiple lines of text. Although REI had received some $50 million from the United States Postal Service to develop this technology for reading and sorting mail, the Postmaster General and other top officials of the Postal Service were urging mailers to use nine-digit zip codes (Zip + 4), which would provide enough routing information on one line of text to allow single-line scanning machines to sort mail automatically by reading just that line.
Besides Moore, who obviously stood to gain financially from the adoption of multiline technology, some Members of Congress and Government research officers had reservations about the Postal Service’s Zip + 4 policy and its intended reliance on single-line readers. Critics maligned single-line scanning technology, objected to the foreign sources of single-line scanners, decried the burden of remembering the four extra numbers, and echoed the conclusion reached by the United States Office of Technology Assessment, that use of the single-line scanners in preference to multiliners would cost the Postal Service $1 million a day in operational losses.
Moore built on this opposition to Zip + 4, by lobbying Members of Congress, testifying before congressional committees, and supporting a “Buy American” rider to the Postal Service’s 1985 appropriations bill. Notwithstanding alleged requests by the Postmaster General to be quiet, REI followed its agenda by hiring a public-relations firm, Gnau and Associates, Inc. (GAI), which one of the Postal Service’s governors, Peter Voss, had recommended.
The campaign succeeded, and in July 1985 the Postal Service made what it called a “mid-course correction” and embraced multiline technology. Brief for Respondent 4. But the change of heart did not extend to Moore and REI, for the Service’s ensuing order of multiline equipment, valued somewhere between $250 million and $400 million went to a competing firm.
Not only did REI lose out on the contract, but Moore and REI were soon entangled in two investigations by Postal Service inspectors. The first looked into the purported payment of kickbacks by GAI to Governor Voss for Voss’s recommendations of GAI’s services, as in the case of REI; the second sought to document REI’s possibly improper role in the search for a new Postmaster General. Notwithstanding very limited evidence linking Moore and REI to any wrongdoing, an Assistant United States Attorney decided to bring criminal charges against them, and in 1988 the grand jury indicted Moore, REI, and REI’s vice president. At the close of the Government’s case, after six weeks of trial, however, the District Court concluded that there was a “complete lack of direct evidence” connecting the defendants to any of the criminal wrongdoing alleged, and it granted the REI defendants’ motion for judgment of acquittal. United States v Recognition Equip. Inc., 725 F. Supp. 587, 596 (DC 1989).
Moore then brought an action in the Northern District of Texas for civil liability under Bivens v Six Unknown Fed. Narcotics Agents, 403 U. S. 388 (1971), against the prosecutor and the five postal inspectors who are petitioners here (a sixth having died). His complaint raised five causes of action, only one of which is relevant here, the claim that the prosecutor and the inspectors had engineered his criminal prosecution in retaliation for criticism of the Postal Service, thus violating the First Amendment. In the course of these proceedings Moore has argued, among other things, that the postal inspectors launched a criminal investigation against him well before they had any inkling of either of the two schemes mentioned above, that the inspectors targeted him for his lobbying activities, and that they pressured the United States Attorney’s Office to have him indicted. Moore also sought recovery from the United States under the Federal Tort Claims Act (FTCA). The District Court dismissed the claims against the Assistant United States Attorney in accordance with the absolute immunity for prosecutorial judgment, and rejected an abuse-of-process claim against the inspectors. Moore v Valder, Civil Action No. 3:91–CV–2491–G (ND Tex., Sept. 21, 1992).
The claims remaining were transferred to the District Court for the District of Columbia, where Moore’s suit was dismissed in its entirety, Civ. Nos. 92–2288 (NHJ), 93–0324 (NHJ), 1993 WL 405785 (Sept. 24, 1993), only to have the Court of Appeals for the District of Columbia Circuit reinstate the retaliatory-prosecution claim. Moore v Valder, 65 F. 3d 189 (1995). The District Court then permitted limited discovery on that matter so far as the inspectors were involved, but again dismissed the remaining charges against the United States and the prosecutor. Moore v Valder, Civil Action No. 92–2288 (NHJ) et al., Record, Tab No. 32 (Memorandum Opinion, Feb. 5, 1998). Although Moore succeeded in having the District of Columbia Circuit reinstate his FTCA claim against the United States, the dismissal of his claims against the prosecutor was affirmed. Moore v United States, 213 F. 3d 705 (2000).
With the remainder of the case back in District Court, the inspectors moved for summary judgment, urging that because the underlying criminal charges were supported by probable cause they were entitled to qualified immunity from a retaliatory-prosecution suit. The District Court denied the motion, and the Court of Appeals affirmed. 388 F. 3d 871 (2004).
The Courts of Appeals have divided on the issue of requiring evidence of a lack of probable cause in 42 U. S. C. §1983 and Bivens retaliatory-prosecution suits. Some Circuits burden plaintiffs with the obligation to show its absence. See, e.g., Wood v Kesler, 323 F. 3d 872, 883 (CA11 2003); Keenan v Tejeda, 290 F. 3d 252, 260 (CA5 2002); Mozzochi v Borden, 959 F. 2d 1174, 1179–1180 (CA2 1992). Others, including the District of Columbia Circuit, impose no such requirement. See, e.g., Poole v County of Otero, 271 F. 3d 955, 961 (CA10 2001); Haynesworth v Miller, 820 F. 2d 1245, 1256–1257 (CADC 1987). We granted certiorari, 545 U. S. ___ (2005), to resolve the Circuit split and now reverse.
Official reprisal for protected speech “offends the Constitution [because] it threatens to inhibit exercise of the protected right,” Crawford-El v Britton, 523 U. S. 574, n. 10 (1998), and the law is settled that as a general matter the First Amendment prohibits government officials from subjecting an individual to retaliatory actions, including criminal prosecutions, for speaking out, id., at 592; see also Perry v Sindermann, 408 U. S. 593, 597 (1972) (noting that the government may not punish a person or deprive him of a benefit on the basis of his “constitutionally protected speech”). Some official actions adverse to such a speaker might well be unexceptionable if taken on other grounds, but when nonretaliatory grounds are in fact insufficient to provoke the adverse consequences, we have held thatretaliation is subject to recovery as the but-for cause of official action offending the Constitution. See Crawford-El, supra, at 593; Mt. Healthy City Bd. of Ed. v Doyle, 429 U. S. 274, 283–284 (1977) (adverse action against government employee cannot be taken if it is in response to the employee’s “exercise of constitutionally protected First Amendment freedoms”). When the vengeful officer is federal, he is subject to an action for damages on the authority of Bivens. See 403 U. S., at 397.
Despite a procedural history portending another Jarndyce v Jarndyce, the issue before us is straightforward: whether a plaintiff in a retaliatory-prosecution action must plead and show the absence of probable cause for pressing the underlying criminal charges.
The inspectors argue on two fronts that absence of probable cause should be an essential element. Without such a requirement, they first say, the Bivens claim is too readily available. A plaintiff can afflict a public officer with disruption and expense by alleging nothing more, in practical terms, than action with a retaliatory animus, a subjective condition too easy to claim and too hard to defend against. Brief for Petitioners 21–23; see also National Archives and Records Admin. v Favish, 541 U. S. 157, 175 (2004) (allegations of government misconduct are “‘easy to allege and hard to disprove’”). In the inspectors’ view, some “objective” burden must be imposed on these plaintiffs, simply to filter out the frivolous. The second argument complements the first, for the inspectors believe that the traditional tort of malicious prosecution tells us what the objective requirement should be. Brief for Petitioners 24–29. In an action for malicious prosecution after an acquittal, a plaintiff must show that the criminal action was begun without probable cause for charging the crime in the first place; the inspectors see retaliatory prosecution under Bivens as a close cousin of malicious prosecution under common law, making the latter’s no-probable-cause requirement a natural feature of the constitutional tort.See Heck v Humphrey, 512 U. S. 477, and 484, n. 4 (1994).
In fact, we think there is a fair argument for what the inspectors call an “objective” fact requirement in this type of case, but the nub of that argument differs from the two they set out, which we will deal with only briefly. As for the invitation to rely on common-law parallels, we certainly are ready to look at the elements of common-law torts when we think about elements of actions for constitutional violations, see Carey v Piphus, 435 U. S. 247, 258 (1978), but the common law is best understood here more as a source of inspired examples than of prefabricated components of Bivens torts. See, e.g., Albright v Oliver, 510 U. S. 266, 277, n. 1 (1994) (Ginsburg, J., concurring); Bivens, supra, at 394; cf. Baker v McCollan, 443 U. S. 137, 146 (1979). And in this instance we could debate whether the closer common-law analog to retaliatory prosecution is malicious prosecution (with its no-probable-cause element) or abuse of process (without it). Compare Heck, 512 U. S., at 483–485, and 484, n. 4 with id., at 493–496 (Souter, J., concurring in judgment).
Nor is there much leverage in the fear that without a filter to screen out claims federal prosecutors and federal courts will be unduly put upon by the volume of litigation. The basic concern is fair enough, but the slate is not blank. Over the past 25 years fewer than two dozen damages actions for retaliatory prosecution under Bivens or §1983 have come squarely before the Federal Courts of Appeals, and there is no disproportion of those cases in Circuits that do not require showing an absence of probable cause.
It is, instead, the need to prove a chain of causation from animus to injury, with details specific to retaliatory-prosecution cases, that provides the strongest justification for the no-probable-cause requirement espoused by the inspectors. Although a Bivens (or §1983) plaintiff must show a causal connection between a defendant’s retaliatory animus and subsequent injury in any sort of retaliation action, see Crawford-El, 523 U. S., at 593; Mt. Healthy, 429 U. S., at 285–287, the need to demonstrate causation in the retaliatory-prosecution context presents an additional difficulty that can be understood by comparing the requisite causation in ordinary retaliation claims, where the government agent allegedly harboring the animus is also the individual allegedly taking the adverse action, with causation in a case like this one.
Take the example of a public employee’s claim that he was fired for speech criticizing the government. See, e.g., Pickering v Board of Ed. of Township High School Dist. 205, Will Cty., 391 U. S. 563, 566–567 (1968) (allegation that a school board dismissed a teacher for writing a public letter critical of the board’s financial administration). While the employee plaintiff obviously must plead and prove adverse official action in retaliation for making the statements, our discussions of the elements of the constitutional tort do not specify any necessary details about proof of a connection between the retaliatory animus and the discharge, which will depend on the circumstances. Cf. Crawford-El, supra, at 593 (“[A]t least with certain types of claims, proof of an improper motive is not sufficient to establish a constitutional violation – there must also be evidence of causation”). The cases have simply taken the evidence of the motive and the discharge as sufficient for a circumstantial demonstration that the one caused the other. See, e.g., Mt. Healthy, supra, at 287; see also Arlington Heights v Metropolitan Housing Development Corp., 429 U. S. 252, 270, n. 21 (1977). It is clear, moreover, that the causation is understood to be but-for causation, without which the adverse action would not have been taken; we say that upon a prima facie showing of retaliatory harm, the burden shifts to the defendant official to demonstrate that even without the impetus to retaliate he would have taken the action complained of (such as firing the employee). See Mt. Healthy, 429 U. S., at 287. If there is a finding that retaliation was not the but-for cause of the discharge, the claim fails for lack of causal connection between unconstitutional motive and resulting harm, despite proof of some retaliatory animus in the official’s mind. See ibid. It may be dishonorable to act with an unconstitutional motive and perhaps in some instances be unlawful, but action colored by some degree of bad motive does not amount to a constitutional tort if that action would have been taken anyway. See Crawford-El, supra, at 593; Mt. Healthy, supra, at 285–286.
When the claimed retaliation for protected conduct is a criminal charge, however, a constitutional tort action will differ from this standard case in two ways. Like any other plaintiff charging official retaliatory action, the plaintiff in a retaliatory-prosecution claim must prove the elements of retaliatory animus as the cause of injury, and the defendant will have the same opportunity to respond to a prima facie case by showing that the action would have been taken anyway, independently of any retaliatory animus. What is different about a prosecution case, however, is that there will always be a distinct body of highly valuable circumstantial evidence available and apt to prove or disprove retaliatory causation, namely evidence showing whether there was or was not probable cause to bring the criminal charge. Demonstrating that there was no probable cause for the underlying criminal charge will tend to reinforce the retaliation evidence and show that retaliation was the but-for basis for instigating the prosecution, while establishing the existence of probable cause will suggest that prosecution would have occurred even without a retaliatory motive. This alone does not mean, of course, that a Bivens or §1983 plaintiff should be required to plead and prove no probable cause, but it does mean that litigating probable cause will be highly likely in any retaliatory-prosecution case, owing to its powerful evidentiary significance.
The second respect in which a retaliatory-prosecution case is different also goes to the causation that a Bivens plaintiff must prove; the difference is that the requisite causation between the defendant’s retaliatory animus and the plaintiff’s injury is usually more complex than it is in other retaliation cases, and the need to show this more complex connection supports a requirement that no probable cause be alleged and proven. A Bivens (or §1983) action for retaliatory prosecution will not be brought against the prosecutor, who is absolutely immune from liability for the decision to prosecute, Imbler v Pachtman, 424 U. S. 409, 431 (1976). Instead, the defendant will be a non-prosecutor, an official, like an inspector here, who may have influenced the prosecutorial decision but did not himself make it, and the cause of action will not be strictly for retaliatory prosecution, but for successful retaliatory inducement to prosecute. The consequence is that a plaintiff like Moore must show that the nonprosecuting official acted in retaliation, and must also show that he induced the prosecutor to bring charges that would not have been initiated without his urging.
Thus, the causal connection required here is not merely between the retaliatory animus of one person and that person’s own injurious action, but between the retaliatory animus of one person and the action of another. See 213 F. 3d, at 710 (“In order to find that a defendant procured a prosecution, the plaintiff must establish ‘a chain of causation’ linking the defendant’s actions with the initiation of criminal proceedings”); see also Barts v Joyner, 865 F. 2d 1187, 1195 (CA11 1989) (plaintiff seeking damages incident to her criminal prosecution would have to show that the police, who allegedly acted in violation of law in securing her arrest, unduly pressured or deceived prosecutors); Dellums v Powell, 566 F. 2d 167, 192–193 (CADC 1977) (where allegation of misconduct is directed at police, a malicious-prosecution claim cannot stand if the decision made by the prosecutor to bring criminal charges was independent of any pressure exerted by police); cf. Smiddy v Varney, 665 F. 2d 261, 267 (CA9 1981) (“[W]here police officers do not act maliciously or with reckless disregard for the rights of an arrested person, they are not liable for damages suffered by the arrested person after a district attorney files charges unless the presumption of independent judgment by the district attorney is rebutted”).
Herein lies the distinct problem of causation in cases like this one. Evidence of an inspector’s animus does not necessarily show that the inspector induced the action of a prosecutor who would not have pressed charges otherwise. Moreover, to the factual difficulty of divining the influence of an investigator or other law enforcement officer upon the prosecutor’s mind, there is an added legal obstacle in the longstanding presumption of regularity accorded to prosecutorial decisionmaking. See Reno v American-Arab Anti-Discrimination Comm., 525 U. S. 471, 489–490 (1999); United States v Armstrong, 517 U. S. 456, 464–466 (1996). And this presumption that a prosecutor has legitimate grounds for the action he takes is one we do not lightly discard, given our position that judicial intrusion into executive discretion of such high order should be minimal, see Wayte v United States, 470 U. S. 598, 607–608 (1985).
Some sort of allegation, then, is needed both to bridge the gap between the nonprosecuting government agent’s motive and the prosecutor’s action, and to address the presumption of prosecutorial regularity. And at the trial stage, some evidence must link the allegedly retaliatory official to a prosecutor whose action has injured the plaintiff. The connection, to be alleged and shown, is the absence of probable cause.
It would be open to us, of course, to give no special prominence to an absence of probable cause in bridging the causal gap, and to address this distinct causation concern at a merely general level, leaving it to such pleading and proof as the circumstances allow. A prosecutor’s disclosure of retaliatory thinking on his part, for example, would be of great significance in addressing the presumption and closing the gap. So would evidence that a prosecutor was nothing but a rubber stamp for his investigative staff or the police. Cf. Mt. Healthy, 429 U. S., at 281–283. (evidence that the board of education, which formally decided not to rehire a teacher, was only nominally distinct from the school superintendent, who allegedly bore the retaliatory animus). In fact, though, these examples are likely to be rare and consequently poor guides in structuring a cause of action. In most cases, for instance, it would be unrealistic to expect a prosecutor to reveal his mind even to the degree that this record discloses, with its reported statement by the prosecutor that he was not galvanized by the merits of the case, but sought the indictment against Moore because he wanted to attract the interest of a law firm looking for a tough trial lawyer.
Accordingly, the significance of probable cause or the lack of it looms large, being a potential feature of every case, with obvious evidentiary value. True, it is not necessarily dispositive: showing an absence of probable cause may not be conclusive that the inducement succeeded, and showing its presence does not guarantee that inducement was not the but-for fact in a prosecutor’s decision. But a retaliatory motive on the part of an official urging prosecution combined with an absence of probable cause supporting the prosecutor’s decision to go forward are reasonable grounds to suspend the presumption of regularity behind the charging decision, see Bordenkircher v Hayes, 434 U. S. 357, 364 (1978) (emphasizing that “so long as the prosecutor has probable cause,” the charging decision is generally discretionary), and enough for a prima facie inference that the unconstitutionally motivated inducement infected the prosecutor’s decision to bring the charge.
Our sense is that the very significance of probable cause means that a requirement to plead and prove its absence will usually be cost free by any incremental reckoning. The issue is so likely to be raised by some party at some point that treating it as important enough to be an element will be a way to address the issue of causation without adding to time or expense. See n. 7, supra. In this case, for example, Moore cannot succeed in the retaliation claim without showing that the Assistant United States Attorney was worse than just an unabashed careerist, and if he can show that the prosecutor had no probable cause, the claim of retaliation will have some vitality.
In sum, the complexity of causation in a claim that prosecution was induced by an official bent on retaliation should be addressed specifically in defining the elements of the tort. Probable cause or its absence will be at least an evidentiary issue in practically all such cases. Because showing an absence of probable cause will have high probative force, and can be made mandatory with little or no added cost, it makes sense to require such a showing as an element of a plaintiff’s case, and we hold that it must be pleaded and proven.
The judgment of the Court of Appeals is reversed, and the case is remanded for further proceedings consistent with this opinion.
It is so ordered.
The Chief Justice and Justice Alito took no part in the consideration or decision of this case.
(with whom Justice Breyer joins, dissenting)
The Court of Appeals, reviewing the record so far made, determined that “[t]he evidence of retaliatory motive [came] close to the proverbial smoking gun.” 388 F. 3d 871, 884 (CADC 2004). The record also indicated that the postal inspectors engaged in “unusual prodding,” strenuously urging a reluctant U. S. Attorney’s Office to press charges against Moore. Ibid. Following Circuit precedent, the Court of Appeals held that “once a plaintiff shows [conduct sheltered by the First Amendment] to have been a motivating factor in the decision to press charges,” the burden shifts to the defending officials to show that the case would have been pursued anyway. Id., at 878.
Recognizing that this case is now directed against the instigating postal inspectors alone, not the prosecutor, I would not assign to the plaintiff the burden of pleading and proving the absence of probable cause for the prosecution. Instead, in agreement with the Court of Appeals, I would assign to the postal inspectors who urged the prosecution the burden of showing that, had there been no retaliatory motive and importuning, the U. S. Attorney’s Office nonetheless would have pursued the case.
Under the Court’s proof burden allocation, which saddles plaintiff – the alleged victim – with the burden to plead and prove lack of probable cause, only entirely “baseless prosecutions” would be checked. Id., at 879. So long as the retaliators present evidence barely sufficient to establish probable cause and persuade a prosecutor to act on their thin information, they could accomplish their mission cost free. Their victim, on the other hand, would incur not only the costs entailed in mounting a defense, he likely would sustain a reputational loss as well, and neither loss would be compensable under federal law. Under the D. C. Circuit’s more speech-protective formulation, “[a] Bivens [v Six Unknown Fed. Narcotics Agents, 403 U. S. 388 (1971),] recovery remains possible .... in those rare cases where strong motive evidence combines with weak probable cause to support a finding that the [investigation and ensuing] prosecution would not have occurred but for the [defending] officials’ retaliatory animus.” Id., at 881. That such situations “are likely to be rare,” it seems to me, does not warrant “structuring a cause of action,” ante, at 14, that precludes relief when they do arise.
For reasons fully developed in the D. C. Circuit’s opinion, I conclude that, in full accord with this Court’s decision in Mt. Healthy City Bd. of Ed. v Doyle, 429 U. S. 274, 287 (1977), the Court of Appeals’ decision strikes the proper balance. I would, therefore, affirm the Circuit’s judgment.
 See, e.g., Seaberry, Durenberger Begins Campaign Against Nine-Digit Zip Code, Washington Post, Feb. 24, 1981, p. E4 (describing Senator David Durenberger’s reference to the Zip + 4 campaign as “‘a mnemonic plague of contagious digititous’”).
 “Bivens established that the victims of a constitutional violation by a federal agent have a right to recover damages against the official in federal court despite the absence of any statute conferring such a right.” Carlson v Green, 446 U. S. 14, 18 (1980). Though more limited in some respects not relevant here, a Bivens action is the federal analog to suits brought against state officials under Rev Stat. §1979, 42 U. S. C. §1983. See Wilson v Layne, 526 U. S. 603, 609 (1999) ; see also Waxman & Morrison, What Kind of Immunity? Federal Officers, State Criminal Law, and the Supremacy Clause, 112 Yale L. J. 2195, 2208 (2003) (“Section 1983 applies .... to state and local officers, [and] the Supreme Court in Bivens .... inferred a parallel damages action against federal officers”).
 Moore and his wife had originally filed this complaint jointly. Her claims were dismissed for lack of standing.
 See 2 C. Dickens, Bleak House 85 (1853).
 Moore contends that we (like the Court of Appeals before us) exceed our appellate jurisdiction when we address the issue of probable cause, see Brief for Respondent 37–39, but his argument is mistaken. It is true that the disagreement over a no-probable-cause requirement arose on the inspectors’ motion for summary judgment on their qualified-immunity defense; Moore stresses that an interlocutory appeal can be taken from the rejection of qualified immunity at the summary-judgment stage only on questions turning on the definition of the violation, not on the sufficiency of the evidence to show that a defendant is in fact entitled to the immunity claimed. See Mitchell v Forsyth, 472 U. S. 511, 528 (1985). Moore says that the issue of probable cause or its absence is simply an evidentiary matter going to entitlement in fact. But the inspectors are making more than a claim about the evidence in this case: they are arguing that we should hold that a showing of no probable cause is an element of the kind of claim Moore is making against them. In agreeing with the inspectors, we are addressing a requirement of causation, which Moore must plead and prove in order to win, and our holding does not go beyond a definition of an element of the tort, directly implicated by the defense of qualified immunity and properly before us on interlocutory appeal. See ibid.; see also Crawford-El v Britton, 523 U. S. 574, 588, 592–593 (1998); Mt. Healthy City Bd. of Ed. v Doyle, 429 U. S. 274, 285–286 (1977) .
 In fact, many of the appellate challenges have been brought in the Second, Fifth, and Eleventh Circuits, all of which require plaintiffs to show an absence of probable cause. See, e.g., Izen v Catalina, 398 F. 3d 363 (CA5 2005) (per curiam); Wood v Kesler, 323 F. 3d 872 (CA11 2003); Keenan v Tejeda, 290 F. 3d 252 (CA5 2002); Singer v Fulton County Sheriff, 63 F. 3d 110 (CA2 1995); Post v Fort Lauderdale, 7 F. 3d 1552 (CA11 1993); Mozzochi v Borden, 959 F. 2d 1174 (CA2 1992); Magnotti v Kuntz, 918 F. 2d 364 (CA2 1990).
 Indeed, even though the Court of Appeals in this case held that plaintiffs do not have to show an absence of probable cause in order to make retaliatory-prosecution claims, it nevertheless acknowledged probable cause’s significance in such suits. See 388 F. 3d 871, 881 (CADC 2004) (“Given that probable cause ordinarily suffices to initiate a prosecution, that showing will be enough in most cases to establish that prosecution would have occurred absent bad intent. A Bivens recovery remains possible, however, in those rare cases where strong motive evidence combines with weak probable cause to support a finding that the prosecution would not have occurred but for the officials’ retaliatory animus”).
 An action could still be brought against a prosecutor for conduct taken in an investigatory capacity, to which absolute immunity does not extend. See Buckley v Fitzsimmons, 509 U. S. 259, 274–276 (1993) (no absolute immunity when prosecutor acts in administrative capacity); Burns v Reed, 500 U. S. 478, 492–495 (1991) (absolute immunity does not attach when a prosecutor offers legal advice to the police regarding interrogation practices). In fact, Moore’s complaint charged the prosecutor with acting in an investigative as well as in a prosecutorial capacity, see App. 45, but dismissal of the complaint as against the prosecutor was affirmed in 213 F. 3d 705, 710 (CADC 2000), and no claim against him is before us now.
 No one here claims that simply conducting a retaliatory investigation with a view to promote a prosecution is a constitutional tort. That is not part of Moore’s complaint. See App. 33–34, 38–45. Whether the expense or other adverse consequences of a retaliatory investigation would ever justify recognizing such an investigation as a distinct constitutional violation is not before us.
 Some may suggest that we should structure a cause of action in the alternative, dispensing with a requirement to show no probable cause when a plaintiff has evidence of a direct admission by a prosecutor that, irrespective of probable cause, the prosecutor’s sole purpose in initiating a criminal prosecution was to acquiesce to the inducements of other government agents, who themselves harbored retaliatory animus. Cf. United States v Armstrong, 517 U. S. 456, n. 3 (1996) (leaving open the question “whether a [criminal] defendant must satisfy the similarly situated requirement in a case ‘involving direct admissions by [prosecutors] of discriminatory purpose’ ” (brackets in original)). But this would seem a little like proposing that retirement plans include the possibility of winning the lottery. Unambiguous admissions of successful inducement are likely to be rare, and hassles over the adequacy of admissions will be the predictable result, if any exemption to a no-probable-cause requirement is allowed.
all rights reserved