PUCKETT v. UNITED STATES, 556 U.S. 129

Justice Scalia delivered the opinion of the Court. The question presented by this case is whether a forfeited claim that the Government has violated the terms of a plea agreement is subject to the plain-error standard of review set forth in Rule 52(

b) of the Federal Rules of Criminal Procedure. I In July 2002, James Puckett was indicted by a grand jury in the Northern District of Texas on one count of armed bank robbery, 18 U. S. C. § 2113(a), (d), and one count of using a firearm during and in relation to a crime of violence, § 924(c)(1). He negotiated a plea agreement with the Government, which was filed with the District Court on September 3, 2003. As part of that deal, Puckett agreed to plead guilty to both counts, waive his trial rights, and cooperate with the Government by being truthful regarding his participation in criminal activities.

App. 51a-53a. In exchange, the Government agreed to the following two terms: “8. The government agrees that Puckett has demonstrated acceptance of responsibility and thereby qualifies for a three-level reduction in his offense level. “9. The government also agrees to request that Puckett’s sentence be placed at the lowest end of the guideline level deemed applicable by the Court.” Id., at 54a. To satisfy the first of these obligations, the Government filed a motion in the District Court pursuant to § 3E1.1 of the United States Sentencing Commission’s Guidelines Manual (Nov. 2003) (USSG).

That provision directs sentencing courts to decrease a defendant’s offense level under the Guidelines by two levels if he “clearly demonstrates acceptance of responsibility for his offense,” and by a third level “upon motion of the government stating that the defendant has assisted authorities in the investigation or prosecution of his own misconduct by timely notifying authorities of his intention to enter a plea of guilty.” Two weeks later, the District Court held a plea colloquy, see Fed. Rule Crim. Proe. 11(b), and accepted Puckett’s plea.

Because of delays due to health problems experienced by Puckett, sentencing did not take place for almost three years. In the interim, Puckett assisted another man in a scheme to defraud the Postal Service, and confessed that assistance (under questioning) to a probation officer. The officer prepared an addendum to Puckett’s presentence report recommending that he receive no §3E1.1 reduction for acceptance of responsibility, on the theory that true acceptance of responsibility requires termination of criminal conduct. See USSG §3E1.1, comment., n. 1(b).

When sentencing finally did take place on May 4, 2006, Puckett’s counsel objected to the addendum, pointing out that the Government had filed a motion requesting that the full three-level reduction in offense level be granted. The District Judge turned to the prosecutor, who responded that the motion was filed “a long time ago,” App. 79a, before Puckett had engaged in the additional criminal behavior. She made clear that the Government opposed any reduction in Puckett’s offense level for acceptance of responsibility.

The probation officer then added his view that under the Guidelines, a reduction would be improper. After hearing these submissions, the District Judge concluded that even assuming he had the discretion to grant the reduction, he would not do so. “[I]t’s so rare [as] to be unknown around here where one has committed a crime subsequent to the crime for which they appear before the court and for them even then to get the three points.” Id., at 80a-81a.

He agreed, however, to follow the recommendation that the Government made, pursuant to its commitment in the plea agreement, that Puckett be sentenced at the low end of the applicable Guidelines range, which turned out to be 262 months in prison for the armed bank robbery and a mandatory minimum consecutive term of 84 months for the firearm crime. Had the District Court granted the three-level reduction for acceptance of responsibility, the bottom of the Guidelines range would have been 188 months for the robbery; the firearm sentence would not have been affected.

Importantly, at no time during the exchange did Puckett’s counsel object that the Government was violating its obligations under the plea agreement by backing away from its request for the reduction. He never cited the relevant provision of the plea agreement. And he did not move to withdraw Puckett’s plea on grounds that the Government had broken its sentencing promises. On appeal to the United States Court of Appeals for the Fifth Circuit, Puckett did argue, inter alia, that the Government violated the plea agreement at sentencing.

The Government conceded that by objecting to the reduction for acceptance of responsibility, it had violated the obligation set forth in paragraph 8 of the agreement, but maintained that Puckett had forfeited this claim by failing to raise it in the District Court. The Court of Appeals agreed, and applied the plain-error standard that Rule 52(

b) makes applicable to unpreserved claims of error. 505 F. 3d 377, 384 (2007). It held that although error had occurred and was obvious, Puckett had not satisfied the third prong of the plain-error analysis by demonstrating that the error affected his substantial rights, i. e., caused him prejudice. Id., at 386. Especially in light of the District Judge’s statement that granting a reduction when the defendant had continued to engage in criminal conduct was “so rare [as] to be unknown,” Puckett could not show that the Government’s breach had affected his ultimate sentence.

The Court of Appeals accordingly affirmed the conviction and sentence. Id., at 388. We granted certiorari, 554 U. S. 945 (2008), to consider a question that has divided the Federal Courts of Appeals: whether Rule 52(b)’s plain-error test applies to a forfeited claim, like Puckett’s, that the Government failed to meet its obligations under a plea agreement. See In re Sealed Case, 356 F. 3d 313, 315-318 (CADC 2004) (discussing conflict among the Circuits). Concluding that Rule 52(

b) does apply and in the usual fashion, we now affirm. II If a litigant believes that an error has occurred (to his detriment) during a federal judicial proceeding, he must object in order to preserve the issue. If he fails to do so in a timely manner, his claim for relief from the error is forfeited. “No procedural principle is more familiar to this Court than that a ... right may be forfeited in criminal as well as civil eases by the failure to make timely assertion of the right before a tribunal having jurisdiction to determine it.” Yakus v. United States, 321 U. S. 414, 444 (1944).

If an error is not properly preserved, appellate-court authority to remedy the error (by reversing the judgment, for example, or ordering a new trial) is strictly circumscribed. There is good reason for this; “anyone familiar with the work of courts understands that errors are a constant in the trial process, that most do not much matter, and that a reflexive inclination by appellate courts to reverse because of unpreserved error would be fatal.” United States v. Padilla, 415 F. 3d 211, 224 (CA1 2005) (Boudin, C. J., concurring).

This limitation on appellate-court authority serves to induce the timely raising of claims and objections, which gives the district court the opportunity to consider and resolve them. That court is ordinarily in the best position to determine the relevant facts and adjudicate the dispute. In the case of an actual or invited procedural error, the district court can often correct or avoid the mistake so that it cannot possibly affect the ultimate outcome.

And of course the contemporaneous-objection rule prevents a litigant from “'sandbagging’” the court — remaining silent about his objection and belatedly raising the error only if the case does not conclude in his favor. Cf. Wainwright v. Sykes, 433 U. S. 72, 89 (1977); see also United States v. Vonn, 535 U. S. 55, 72 (2002). In federal criminal cases, Rule 51(

b) tells parties how to preserve claims of error: “by informing the court — when the court ruling or order is made or sought — of the action the party wishes the court to take, or the party’s objection to the court’s action and the grounds for that objection.” Failure to abide by this contemporaneous-objection rule ordinarily precludes the raising on appeal of the unpreserved claim of trial error. See United States v. Young, 470 U. S. 1, 15, and n. 12 (1985). Rule 52(b), however, recognizes a limited exception to that preclusion.

The Rule provides, in full: “A plain error that affects substantial rights may be considered even though it was not brought to the court’s attention.” We explained in United States v. Olano, 507 U. S. 725 (1993), that Rule 52(

b) review — so-called “plain-error review” — involves four steps, or prongs. First, there must be an error or defect — some sort of “[deviation from a legal rule” — that has not been intentionally relinquished or abandoned, i. e., affirmatively waived, by the appellant. Id., at 732-733. Second, the legal error must be clear or obvious, rather than subject to reasonable dispute. See id., at 734. Third, the error must have affected the appellant’s substantial rights, which in the ordinary case means he must demonstrate that it “affected the outcome of the district court proceedings.” Ibid.

Fourth and finally, if the above three prongs are satisfied, the court of appeals has the discretion to remedy the error — discretion which ought to be exercised only if the error “ ‘seriously affect[s] the fairness, integrity or public reputation of judicial proceedings.’” Id., at 736 (quoting United States v. Atkinson, 297 U. S. 157, 160 (1936)). Meeting all four prongs is difficult, “as it should be.” United States v. Dominguez Benitez, 542 U. S. 74, 83, n. 9 (2004). We have repeatedly cautioned that “[a]ny unwarranted extension” of the authority granted by Rule 52(

b) would disturb the careful balance it strikes between judicial efficiency and the redress of injustice, see Young, supra, at 15; and that the creation of an unjustified exception to the Rule would be “[e]ven less appropriate,” Johnson v. United States, 520 U. S. 461, 466 (1997). The real question in this ease is not whether plain-error review applies when a defendant fails to preserve a claim that the Government defaulted on its plea-agreement obligations, but rather what conceivable reason exists for disregarding its evident application.

Such a breach is undoubtedly a violation of the defendant’s rights, see Santobello v. New York, 404 U. S. 257, 262 (1971), but the defendant has the opportunity to seek vindication of those rights in district court; if he fails to do so, Rule 52(

b) as clearly sets forth the consequences for that forfeiture as it does for all others. Ill Puckett puts forward several possible reasons why plain-error review should not apply in the present context. We understand him to be making effectively four distinct arguments: two doctrinal, two practical. We consider each set in turn.

A Puckett’s primary precedent-based argument proceeds as follows: When the Government breaks a promise that was made to a defendant in the course of securing a guilty plea, the knowing and voluntary character of that plea retroactively vanishes, because (as it turns out) the defendant was not aware of its true consequences. Since guilty pleas must be knowing and voluntary to be valid, McCarthy v. United States, 394 U. S. 459, 466 (1969), the guilty plea is thus void, along with the defendant’s corresponding waiver of his right to trial.

And because, under this Court’s precedents, a waiver of the right to trial must be made by the defendant personally, see Taylor v. Illinois, 484 U. S. 400, 417-418, and n. 24 (1988), no action by counsel alone could resurrect the voided waiver. Therefore, Puckett concludes, counsel’s failure timely to object to a Government breach can have no effect on the analysis, and the court of appeals must always correct the error. This elaborate analysis suffers from at least two defects.

First, there is nothing to support the proposition that the Government’s breach of a plea agreement retroactively causes the defendant’s agreement to have been unknowing or involuntary. Any more than there is anything to support the proposition that a mere breach of contract retroactively causes the other party’s promise to have been coerced or induced by fraud. Although the analogy may not hold in all respects, plea bargains are essentially contracts. See Mabry v. Johnson, 467 U. S. 504, 508 (1984).

When the consideration for a contract fails — that is, when one of the exchanged promises is not kept — we do not say that the voluntary bilateral consent to the contract never existed, so that it is automatically and utterly void; we say that the contract was broken. See 23 R. Lord, Williston on Contracts §63.1 (4th ed. 2002) (hereinafter Williston). The party injured by the breach will generally be entitled to some remedy, which might include the right to rescind the contract entirely, see 26 id., § 68.1 (4th ed. 2003); but that is not the same thing as saying the contract was never validly concluded.

So too here. When a defendant agrees to a plea bargain, the Government takes on certain obligations. If those obligations are not met, the defendant is entitled to seek a remedy, which might in some cases be rescission of the agreement, allowing him to take back the consideration he has furnished, i. e., to withdraw his plea. But rescission is not the only possible remedy; in Santobello we allowed for a re-sentencing at which the Government would fully comply with the agreement — in effect, specific performance of the contract. 404 U. S., at 263.

In any case, it is entirely clear that a breach does not cause the guilty plea, when entered, to have been unknowing or involuntary. It is precisely because the plea was knowing and voluntary (and hence valid) that the Government is obligated to uphold its side of the bargain. Moreover, and perhaps more fundamentally, Puckett’s argument confuses the concepts of waiver and forfeiture. Nobody contends that Puckett’s counsel has waived — that is, intentionally relinquished or abandoned, Olano, 507 U.

S., at 733 — Puckett’s right to seek relief from the Government’s breach. (If he had, there would be no error at all and plain-error analysis would add nothing.) The objection is rather that Puckett forfeited the claim of error through his counsel’s failure to raise the argument in the District Court. This Court’s precedents requiring that certain waivers be personal, knowing, and voluntary are thus simply irrelevant. Those holdings determine whether error occurred, but say nothing about the proper standard of review when the claim of error is not preserved.

The question presented by this case assumes error; only the standard of review is in dispute. Puckett’s second doctrinal attack rests on our decision in Santobello. In that case, the State had promised in a plea deal that it would make no sentencing recommendation, but the prosecutor (apparently unaware of that commitment) asked the state trial court to impose the maximum penalty of one year. Defense counsel immediately objected. 404 U. S., at 259. The trial judge proceeded anyway to impose the 1-year sentence, reassuring Santobello that the prosecutor’s recommendation did not affect his decision.

Id., at 259-260. This Court vacated the conviction and remanded the case because “the interests of justice” would thus be best served. Id., at 262. Puckett maintains that if the “interests of justice” required a remand in Santobello even though the breach there was likely harmless, those same interests call for a remand whenever the Government reneges on a plea bargain, forfeiture or not. We do not agree. Whether an error can be found harmless is simply a different question from whether it can be subjected to plain-error review.

Santobello (given that the error in that case was preserved) necessarily addressed only the former. B Doctrine and precedent aside, Puckett argues that practical considerations counsel against subjecting plea-breach claims to the rule of plain-error review. Specifically, he contends that no purpose would be served by applying the rule; and that plea breaches will always satisfy its four prongs, making its application superfluous. Accepting, arguendo (and dubitante), that policy concerns can ever authorize a departure from the Federal Rules, both arguments are wrong.

Puckett suggests that once the prosecution has broken its agreement, e. g., by requesting a higher sentence than agreed upon, it is too late to “unring” the bell even if an objection is made: The district judge has already heard the request, and under Santobello it does not matter if he was influenced by it. So why demand the futile objection? For one thing, requiring the objection means the defendant cannot “game” the system, “waiting] to see if the sentence later strikes] him as satisfactory,” Vonn, 535 U. S., at 73, and then seeking a second bite at the apple by raising the claim.

For another, the breach itself will not always be conceded. In such a case, the district court if apprised of the claim will be in a position to adjudicate the matter in the first instance, creating a factual record and facilitating appellate review. Thirdly, some breaches may be curable upon timely objection — for example, where the prosecution simply forgot its commitment and is willing to adhere to the agreement.

And finally, if the breach is established but cannot be cured, the district court can grant an immediate remedy (e. g., withdrawal of the plea or resentencing before a different judge) and thus avoid the delay and expense of a full appeal. Puckett also contends that plain-error review “does no substantive work” in the context of the Government’s breach of a plea agreement. Brief for Petitioner 22.

He claims that the third prong, the prejudice prong, has no application, since plea-breach claims fall within “a special category of forfeited errors that can be corrected regardless of their effect on the outcome.” Olano, 507 U. S., at 735. This Court has several times declined to resolve whether “structural” errors — those that affect “the framework within which the trial proceeds,” Arizona v. Fulminante, 499 U. S. 279, 310 (1991) — automatically satisfy the third prong of the plain-error test. Olano, supra, at 735; Johnson, 520 U. S., at 469; United States v. Cotton, 535 U. S. 625, 632 (2002).

Once again we need not answer that question, because breach of a plea deal is not a “structural” error as we have used that term. We have never described it as such, see Johnson, supra, at 468-469, and it shares no common features with errors we have held structural. A plea breach does not “necessarily render a criminal trial fundamentally unfair or an unreliable vehicle for determining guilt or innocence,” Neder v. United States, 527 U.

S. 1, 9 (1999) (emphasis deleted); it does not “defy analysis by 'harmless-error’ standards” by affecting the entire adjudicatory framework, Fulminante, supra, at 309; and the “difficulty of assessing the effect of the error,” United States v. Gonzalez-Lopez, 548 U. S. 140, 149, n. 4 (2006), is no greater with respect to plea breaches at sentencing than with respect to other procedural errors at sentencing, which are routinely subject to harmlessness review, see, e. g., United States v. Teague, 469 F. 3d 205, 209-210 (CA1 2006).

Santobello did hold that automatic reversal is warranted when objection to the Government’s breach of a plea agreement has been preserved, but that holding rested not upon the premise that plea-breach errors are (like “structural” errors) somehow not susceptible, or not amenable, to review for harmlessness, but rather upon a policy interest in establishing the trust between defendants and prosecutors that is necessary to sustain plea bargaining — an “essential” and “highly desirable” part of the criminal process, 404 U. S., at 261-262.

But the rule of contemporaneous objection is equally essential and desirable, and when the two collide we see no need to relieve the defendant of his usual burden of showing prejudice. See Olano, supra, at 734. The defendant whose plea agreement has been broken by the Government will not always be able to show prejudice, either because he obtained the benefits contemplated by the deal anyway (e. g., the sentence that the prosecutor promised to request) or because he likely would not have obtained those benefits in any event (as is seemingly the case here).

On the dissent’s view, a defendant in Puckett’s position has always suffered an impairment of his “substantial rights” under Olano’s third prong, because he has been convicted “in the absence of trial or compliance with the terms of the plea agreement dispensing with the Government’s obligation to prove its case.” Post, at 143-144 (opinion of Souter, J.). But that is simply an ipse dixit recasting the conceded error — breach of the plea agreement — as the effect on substantial rights.

Any trial error can be said to impair substantial rights if the harm is defined as “being convicted at a trial tainted with [fill-in-the-blank] error.” Nor does the fact that there is a “protected liberty interest” at stake render this case different, see post, at 145. That interest is always at stake in criminal cases. Eliminating the third plain-error prong through semantics makes a nullity of Olano’s instruction that a defendant normally “must make a specific showing of prejudice” in order to obtain relief, 507 U. S., at 735.

Puckett contends that the fourth prong of plain-error review likewise has no application because every breach of a plea agreement will constitute a miscarriage of justice. That is not so. The fourth prong is meant to be applied on a case-specific and fact-intensive basis. We have emphasized that a “per se approach to plain-error review is flawed.” Young, 470 U. S., at 17, n. 14. It is true enough that when the Government reneges on a plea deal, the integrity of the system may be called into question, but there may well be countervailing factors in particular cases.

Puckett is again a good example: Given that he obviously did not cease his life of crime, receipt of a sentencing reduction for acceptance of responsibility would have been so ludicrous as itself to compromise the public reputation of judicial proceedings. Of course the second prong of plain-error review also will often have some “bite” in plea-agreement cases. Not all breaches will be clear or obvious. Plea agreements are not always models of draftsmanship, so the scope of the Government’s commitments will on occasion be open to doubt.

Moreover, the Government will often have a colorable (albeit ultimately inadequate) excuse for its nonperformance. See n. 2, supra. * * * Application of plain-error review in the present context is consistent with our cases, serves worthy purposes, has meaningful effects, and is in any event compelled by the Federal Rules. While we recognize that the Government’s breach of a plea agreement is a serious matter, “the seriousness of the error claimed does not remove consideration of it from the ambit of the Federal Rules of Criminal Procedure.” Johnson, 520 U. S., at 466.

The judgment of the Court of Appeals is Affirmed. Justice Souter, with whom Justice Stevens joins, dissenting. Petitioner’s situation does not excite sympathy, but the Court’s holding will stand for a rule in circumstances less peculiar than those here. I disagree with my colleagues with respect to the interest at stake for a criminal defendant in a case like this, and I respectfully dissent.

This case turns on whether plain-error review applies to an unpreserved claim that the Government breached its plea agreement and on identifying the relevant effect, or substantial rights implicated, under the third prong of United States v. Olano, 507 U. S. 725, 734 (1993). I agree with the majority that plain error is the proper test, but depart from the Court’s holding that the effect in question is length of incarceration for the offense charged (as to which the error here probably made no ultimate difference).

I would hold that the relevant effect is conviction in the absence of trial or compliance with the terms of the plea agreement dispensing with the Government’s obligation to prove its case. The first two conditions for recognizing plain error, that there be error and that it be clear, see id., at 732-734, are without doubt satisfied here. Before sentencing, a colloquy in accordance with Federal Rule of Criminal Procedure 11 laid the ground for satisfying the requirement that the error be obvious, by making a public record of the terms of the plea agreement between Puckett and the Government.

Both the written agreement and the Government’s representation to the District Court included the Government’s statement that Puckett qualified for a three-level reduction in his offense level under the Sentencing Guidelines, because of his acceptance of responsibility for his offense. See App. 54a (“The government agrees that Puckett has demonstrated acceptance of responsibility and thereby qualifies for a three-level reduction in his offense level”); id., at 68a (“The government agrees that Mr.

Puckett has demonstrated acceptance of responsibility and thereby would qualify for a three level reduction in his offense level”). Puckett does indeed appear to have satisfied the conditions on which the Government’s commitment was premised: he accepted responsibility for committing “his offense[s]” and “assisted authorities in the investigation or prosecution of his own misconduct by timely notifying authorities of his intention to enter a plea of guilty.” United States Sentencing Commission, Guidelines Manual §3E1.1 (Nov. 2003).

His subsequent criminality (during the unusual 3-year break between his guilty plea and sentencing) was not a -failure on his part to accept responsibility for his prior crimes (the benefit of which the Government had already received by the time Puckett pleaded guilty). In any case, the Government could have insisted on a provision in the plea agreement allowing it to back out of its commitment if Puckett engaged in additional criminal conduct prior to sentencing, and did not do so. It should therefore be bound by the terms of the agreement it made, whether or not Puckett was in fact entitled to the reduction.

In administering the criminal law no less than the civil, parties are routinely bound by agreements they wish they had not made. This is why the Government has no choice but to admit that it breached the plea agreement when, at sentencing, it objected to the three-level reduction. Despite its contention that the plain-error doctrine does not save Puckett from his failure to object at the sentencing hearing, the Government does not deny that a deal is a deal and it does not deny that it broke its word.

The plain-error doctrine will not, however, avail Puckett anything unless the remaining conditions set out in Olano are satisfied, the third requiring a showing that sentencing Puckett on a plea given in return for an unfulfilled promise by the Government violated his substantial rights. See 507 U. S., at 734. The majority understands the effect in question to be length of incarceration. See ante, at 142, n. 4 (“When the rights acquired by the defendant relate to sentencing, the ‘ “outcome” ’ he must show to have been affected is his sentence”).

Since Puckett can hardly show that a court apprised of his subsequent criminality would have given him the three-level reduction even in the absence of the Government’s breach, in the majority’s view he cannot satisfy the “substantial rights” criterion and so fails to qualify for correction of the admitted clear error. I, on the contrary, would identify the effect on substantial rights as the criminal conviction itself, regardless of length of incarceration.

My reason is simply that under the Constitution the protected liberty interest in freedom from eriminal taint, subject to the Fifth Amendment’s due process guarantee of fundamental fairness, is properly understood to require a trial or plea agreement honored by the Government before the stigma of a conviction can be imposed. That protection does not vanish if a convicted defendant turns out to get a light sentence.

It is the trial leading to possible conviction, not the sentencing hearing alone, that is the focus of this guarantee, and it is the possibility of criminal conviction itself, without more, that calls for due process protection. In a legal system constituted this way, it is hard to imagine anything less fair than branding someone a criminal not because he was tried and convicted, but because he entered a plea of guilty induced by an agreement the Government refuses to honor.

Agreements must therefore be kept by the Government as well as by the individual, and if the plain-error doctrine can ever rescue a defendant from the consequence of forfeiting rights by inattention, it should be used when the Government has induced an admission of criminality by making an agreement that it deliberately breaks after the defendant has satisfied his end of the bargain. Redressing such fundamentally unfair behavior by the Government, whether by vacating the plea or enforcing the plea agreement, see Santobello v. New York, 404 U.

S. 257, 263 (1971), is worth the undoubted risk of allowing a defendant to game the system and the additional administrative burdens, see ante, at 134, 140. If the Judiciary is worried about gamesmanship and extra proceedings, all it needs to do is to minimize their likelihood by making it plain that it will require the Government to keep its word or seek rescission of the plea agreement if it has cause to do so.

Thus, I would find that a defendant’s substantial rights have been violated whenever the Government breaches a plea agreement, unless the defendant got just what he bargained for anyway from the sentencing court. What I have said about the third Olano criterion determines my treatment of the fourth, addressing whether leaving the error uncorrected “may be said . . . ‘seriously [to] affect the fairness, integrity or public reputation of judicial proceedings.’ ” 507 U. S., at 744 (Stevens, J., dissenting) (quoting United States v. Atkinson, 297 U. S. 157, 160 (1936)).

If I am right that in this case the protected interest is in the guarantee that no one is liable to spend a day behind bars as a convict without a trial or his own agreement, then the fairness and integrity of the Judicial Branch suffer when a court imprisons a defendant after he pleaded guilty in reliance on a plea agreement, only to have the Government repudiate the obligation it agreed upon. That is precisely what happened here, yet the Judiciary denies relief under an appellate procedure for correcting patent error.

Judicial repute does not escape without damage in the eyes of anyone who sees beyond the oddity of this case. Puckett is entitled to relief because he and every other defendant who may make an agreement with the Government are entitled to take the Government at its word. Puckett insists that the Government keep its word, and if we are going to have a plain-error doctrine at all, the Judiciary has no excuse for closing this generally available avenue of redress to Puckett or to any other criminal defendant standing in his shoes. Puckett points out that in Brady v. United States, 397 U.

S. 742 (1970), we quoted approvingly the Fifth Circuit’s statement that guilty pleas must stand unless induced by “misrepresentation (including unfulfilled or unfulfillable promises),” id., at 755 (quoting Shelton v. United States, 246 F. 2d 571, 572, n. 2 (CA5 1957) (en banc); internal quotation marks omitted). But it is hornbook law that misrepresentation requires an intent at the time of contracting not to perform. 26 Williston § 69.11. It is more difficult to explain the other precedent relied upon by Puckett — our suggestion in Mabry v. Johnson, 467 U.

S. 504, 509 (1984), that “when the prosecution breaches its promise with respect to an executed plea agreement, the defendant pleads guilty on a false premise, and hence his conviction cannot stand.” That statement, like the one in Brady, was dictum. Its conclusion that the conviction cannot stand is only sometimes true (if that is the remedy the court prescribes for the breach). And even when the conviction is overturned, the reason is not that the guilty plea was unknowing or involuntary. We disavow any aspect of the Mabry dictum that contradicts our holding today.

Indeed, in this case the Government might well have argued that it was excused from its obligation to assert “demonstrated acceptance of responsibility” because Puckett’s ongoing criminal conduct hindered performance. See 13 Williston §39.3 (4th ed. 2000). That argument might have convinced us had it been pressed, but the Government conceded the breach, and we analyze the case as it comes to us. We need not confront today the question whether Santobello’s automatic-reversal rule has survived our recent elaboration of harmless-error principles in such cases as Fulminante and Neder.

Because, as we have explained, the breach consists of a wrongful denial of the rights obtained by the defendant through the plea agreement and does not automatically invalidate the plea, we agree with the Government that the question with regard to prejudice is not whether Puckett would have entered the plea had he known about the future violation. Cf. United States v. Dominguez Benitez, 542 U. S. 74, 83 (2004). When the rights acquired by the defendant relate to sentencing, the “ ‘outcome* ” he must show to have been affected is his sentence.