with whom Justice Souter joins, dissenting.
The Court today transforms a defendant’s presence at trial from a Sixth Amendment right into an automatic burden on his credibility. I dissent from the Court’s disposition. In *77Griffin v. California, 880 U. S. 609 (1965), we held that a defendant’s refusal to testify at trial may not be used as evidence of his guilt. In Doyle v. Ohio, 426 U. S. 610 (1976), we held that a defendant’s silence after receiving Miranda warnings did not warrant a prosecutor’s attack on his credibility. Both decisions stem from the principle that where the exercise of constitutional rights is “insolubly ambiguous” as between innocence and guilt, id., at 617, a prosecutor may not unfairly encumber those rights by urging the jury to construe the ambiguity against the defendant.
same principle should decide this case. Ray Agard attended his trial, as was his constitutional right and his statutory duty, and he testified in a manner consistent with other evidence in the case. One evident explanation for the coherence of his testimony cannot be ruled out: Agard may have been telling the truth. It is no more possible to know whether Agard used his presence at trial to figure out how to tell potent lies from the witness stand than it is to know whether an accused who remains silent had no exculpatory story to tell.
imposes on the exercise of Sixth Amendment rights is justified, the Court maintains, because “the central function of the trial... is to discover the truth.” See ante, at 73. A trial ideally is a search for the truth, but I do not agree that the Court’s decision advances that search. The generic accusation that today’s decision permits the prosecutor to make on summation does not serve to distinguish guilty defendants from innocent ones. Every criminal defendant, guilty or not, has the right to attend his trial. U. S. Const., Arndt. 6. Indeed, as the Court grants, ante, at 74, New York law requires defendants to be present when tried. It follows that every defendant who testifies is equally susceptible to a generic accusation about his opportunity for tailoring. The prosecutorial comment at issue, tied only to the defendant’s presence in the courtroom and not to his actual testimony, tarnishes the innocent no *78less than the guilty. Nor can a jury measure a defendant’s credibility by evaluating the defendant’s response to the accusation, for the broadside is fired after the defense has submitted its case. An irrebuttable observation that can be made about any testifying defendant cannot sort those who tailor their testimony from those who do not, much less the guilty from the innocent.
I
The Court of Appeals took a carefully restrained and moderate position in this ease. It held that a prosecutor may not, as part of her summation, use the mere fact of a defendant’s presence at his trial as the basis for impugning his credibility. A prosecutor who wishes at any stage of a trial to accuse a defendant of tailoring specific elements of his testimony to fit with particular testimony given by other witnesses would, under the decision of the Court of Appeals, have leave to do so. See 159 P. 3d 98, 99 (CA2 1998). Moreover, on cross-examination, a prosecutor would be free to challenge a defendant’s overall credibility by pointing out that the defendant had the opportunity to tailor his testimony in general, even if the prosecutor could point to no facts suggesting that the defendant had actually engaged in tailoring. See 117 F. 3d 696, 708, n. 6 (CA2 1997). The Court of Appeals held only that the prosecutor may not launch a general accusation of tailoring on summation. See id., at 709; see also United States v. Chacko, 169 F. 3d 140, 150 (CA2 1999). Thus, the decision below would rein in a prosecutor solely in situations where there is no particular reason to believe that tailoring has occurred and where the defendant has no opportunity to rebut the accusation.
The Court of Appeals’ judgment was correct in light of Griffin and Doy le. Those decisions instruct that when a defendant’s exercise of a constitutional fair trial right is “insolubly ambiguous” as between innocence and guilt, the prosecutor may not urge the jury to construe the bare invocation of the right against the defendant. See Doyle, 426 U. S., at *79617. To be sure, defendants are not categorically exempt from some costs associated with the assertion of their constitutional prerogatives. The Court is correct to say that the truth-seeking function of trials places demands on defendants. In a proper ease, that central function could justify a particular burden on the exercise of Sixth Amendment rights. But the interests of truth are not advanced by allowing a prosecutor, at a time when the defendant cannot respond, to invite the jury to convict on the basis of conduct as consistent with innocence as with guilt. Where burdening a constitutional right will not yield a compensating benefit, as in the present case, there is no justification for imposing the burden.
trials may be served by permitting prosecutors to make accusations of tailoring — even wholly generic accusations of tailoring — as part of cross-examination. Some defendants no doubt do give false testimony calculated to fit with the testimony they hear from other witnesses. If accused on cross-examination of having tailored their testimony, those defendants might display signals of untrustworthiness that it is the province of the jury to detect and interpret. But when a generic argument is offered on summation, it cannot in the slightest degree distinguish the guilty from the innocent. It undermines all defendants equally and therefore does not help answer the question that is the essence of a trial’s search for truth: Is this particular defendant lying to cover his guilt or truthfully narrating his innocence?1
*80In addition to its incapacity to serve the individualized truth-finding function of trials, a generic tailoring argument launched on summation entails the simple unfairness of preventing a defendant from answering the charge. This problem was especially pronounced in the instant case. Under New York law, defendants generally may not bolster their own credibility by introducing their prior consistent statements but may introduce such statements to rebut claims of recent fabrication. See People v. McDaniel, 81 N. Y. 2d 10, 16, 611 N. E. 2d 265, 268 (1993); 117 F. 3d, at 715 (Winter, C. J., concurring). Had the prosecution made its tailoring accusations on cross-examination, Agard might have been able to prove that his story at trial was the same as it had been before he heard the testimony of other witnesses. A prosecutor who can withhold a tailoring accusation until summation can avert such a rebuttal.
The Court’s only support for its choice to ignore the distinction between summation and cross-examination is Reagan v. United States, 157 U. S. 301 (1895), a decision which, by its very terms, does not bear on today’s constitutional controversy. It is true, as the Court says, that Reagan upheld a trial judge’s instruction that questioned the credibility of a testifying defendant in a generic manner, and it is also true that a defendant is no more able to respond to an instruction than to a prosecutor’s summation. But Reagan has no force as precedent for this case because, in the 1895 Court’s view, the instruction there at issue did not burden any constitutional right of the defendant.
The trial court in Reagan instructed the jury that when it evaluated the credibility of the defendant’s testimony, it could consider that defendants have a powerful interest in being acquitted, powerful enough that it might induce some people to lie. See id., at 304-305. This instruction burdened the defendant’s right to testify at his own trial. But the Court that decided Reagan conceived of that right as one dependent on a statute, not on any constitutional prescrip*81tion. See id., at 304 (defendant was qualified to testify under oath pursuant to an 1878 Act of Congress, ch. 37, 20 Stat. 30, which removed the common-law disability that had previously prevented defendants from giving sworn testimony). No one in that 19th-century case suggested that the trial court’s comment exacted a penalty for the exercise of any constitutional right.2 It is thus inaccurate for the Court to portray Reagan as precedent for the proposition that the difference between summation and cross-examination “is not a constitutionally significant distinction.” Ante, at 72. Reagan made no determination of constitutional significance or insignificance, for it addressed no constitutional question.
to bring Reagan within constitutional territory by yoking it to Griffin. The Court asserts that Griffin relied on the very statute that defined the rights of the defendant in Reagan and that Griffin’s holding makes sense only if the statute in Reagan carries constitutional implications. Ante, at 72, n. 3. This argument is flawed in its premise, because Griffin rested solidly on the Fifth Amendment. The Court in Griffin did refer to the 1878 statute at issue in Reagan, but it did so only in connection with its discussion of Wilson v. United States, 149 U. S. 60 (1893), a decision construing a different provision of that statute to prohibit federal prosecutors from commenting to juries on defendants’ failure to testify. See Griffin, 380 U. S., at 612-613. The statute at issue in Reagan and Wilson, now codified at 18 U. S. C. § 3481, provides that defendants in criminal trials have both the right to testify and the right not *82to testify. Reagan concerned the former right, Wilson the latter right, and Griffin the constitutional analogue to the latter right. If the Court in Griffin had regarded the statute as settling the meaning of the Fifth Amendment — an odd position to imagine the Court taking — then it could have rested on Wilson. It did not. It said that Wilson would govern were the question presented a statutory one, but that the question before it was constitutional: “The question remains whether, statute or not, the comment. . . violates the Fifth Amendment.” 380 U. S., at 613 (emphasis added). Thus, the question in Griffin was not controlled by Wilson precisely because the statute construed in Wilson and Reagan was just that — a statute — and not a provision of the Constitution. Accordingly, Griffin provides no support for the Court’s unorthodox- contention that Reagan’s statutory holding was actually of constitutional dimension.3
II
The Court offers two arguments in support of its conclusion that a prosecutor may make the generic tailoring accusations at issue in this case. First, it suggests that such comment has historically not been seen as problematic. *83respondent Agard’s case is readily distinguishable from Griffin. The Court’s historical excur-sus does not even begin to prove that comments like those in this case have ever been accepted as constitutional, and the attempt to distinguish Griffin relies on implausible premises that this Court has previously rejected.
narrative proceeds as follows: In the early days of the Republic, prosecutors had no “need” to suggest that defendants might use their presence at trial to tailor their testimony, because defendants’ (unsworn) statements at trial could be compared with pretrial statements that defendants gave as a matter of course. Later, some States instituted rules requiring defendants to testify before the other witnesses did,4 thus obviating once again any .need to make arguments about tailoring. There is no evidence, the Court says, that any State ever prohibited the kind of generic argument now at issue until recent times.5 So it must be the case that generic tailoring arguments have traditionally been thought unproblematic. Ante, at 65-66.
*84I do not comprehend why the Court finds in this account any demonstration that the prosecutorial comment at issue here has a long history of unchallenged use. If prosecutors in times past had no need to make generic tailoring arguments, it is likely such arguments simply were not made. Notably, the Court calls up no instance of an 18th- or 19th-century prosecutor’s urging that a defendant’s presence at trial facilitated tailored testimony. And if prosecutors did not make such arguments, courts had no occasion to rule them out of order. The absence of old cases prohibiting the comment that the Court now confronts thus scarcely indicates that generic accusations of tailoring have long been considered constitutional.
The Court’s discussion of Griffin is equally unconvincing. The Court posits that a ban on inviting juries to draw adverse inferences from a defendant’s silence differs materially from a ban on inviting juries to draw adverse inferences from a defendant’s presence, because the inference from silence “is not . . . ‘natural or irresistible.’” See ante, at 67 (quoting Griffin, 380 U. S., at 615) (emphasis added by majority). This is a startling statement. It fails to convey what the Court actually said in Griffin, which was that the inference from silence to guilt is “not always so natural or irresistible.” See ibid, (emphasis added). The statement that an inference is not always natural or irresistible implies that the inference is indeed natural or irresistible in many, perhaps most, cases. And so it is. See Mitchell v. United States, 526 U. S. 314, 332 (1999) (Scalia, J., dissenting) (The Griffin rule “runs exactly counter to normal evidentiary inferences: If I ask my son whether he saw a movie I had forbidden him to watch, and he remains silent, the import of his silence is clear.”); Lakeside v. Oregon, 435 U. S. 333, 340 (1978) (It is “very doubtful” that jurors, left to their own devices, would not draw adverse inferences from a defendant’s failure to testify.). It is precisely because the inference is often natural (but nonetheless prohibited) that the jury, if a defendant so *85requests, is instructed not to draw it. Carter v. Kentucky, 450 U. S. 288, 301-303 (1981) (An uninstructed jury is likely to draw adverse inferences from a defendant’s failure to testify, so defendants are entitled to have trial courts instruct juries that no such inference may be drawn.).
inference involved in Griffin is at least as “natural” or “irresistible” as the inference the prosecutor in Agard’s ease invited the jury to draw. There are, to be sure, reasons, why an innocent defendant might not want to testify. Perhaps he fears that his convictions for prior crimes will generate prejudice against him if placed before the jury; perhaps he has an unappealing countenance that could produce the same effect; perhaps he worries that cross-examination will drag into public view prior conduct that, though not unlawful, is deeply embarrassing. For similar reasons, an innocent person might choose to remain silent after arrest. But in either the Griffin scenario of silence at trial or the Doyle scenario of silence after arrest, something beyond the simple innocence of the defendant must be hypothesized in order to explain the defendant’s behavior.
Not so in the present case. If a defendant appears at trial and gives testimony that fits the rest of the evidence, sheer innocence could explain his behavior completely. The inference from silence to guilt in Griffin or from silence to un-trustworthiness in Doyle is thus more direct than the inference from presence to tailoring.6 Unless one has prejudged *86the defendant as guilty, or unless there are specific reasons to believe that particular testimony has been altered, the possibility that the defendant is telling the truth is surely as good an explanation for the coherence of the defendant’s testimony as any that involves wrongful tailoring. I therefore disagree with the Court’s assertion, ante, at 68, that the Court of Appeals’ decision in Agard’s case differs from our decision in Griffin by “requiring] the jury to do what is practically impossible.”7 It makes little sense to maintain that juries able to avoid drawing adverse inferences from a defendant’s silence would be unable to avoid thinking that only a defendant’s opportunity to spin a web of lies could explain the seamlessness of his testimony.
The Court states in the alternative that if proscribing generic accusations of tailoring at summation does not require the jury to do the impossible, then it prohibits prosecutors from “inviting the jury to do what the jury is perfectly entitled to do.” Ante, at 68. The Court offers no prior authority, however, for the proposition that a jury may constitutionally draw the inference now at issue. The Second Circuit thought the matter open, and understandably so in light of Griffin and Carter. But even if juries were permitted to draw the inference in question, it would not follow that prosecutors could urge juries to draw it. Doyle prohibits prosecutors from urging juries to draw adverse inferences from a defendant’s choice to remain silent after re*87ceiving Miranda warnings, but the Court today shows no readiness to say that juries may not draw that inference themselves. See ante, at 74-75. It therefore seems unproblematic to hold that a prosecutor’s latitude for argument is narrower than a jury’s latitude for assessment.
endeavor to distinguish the two inferences, the Court maintains that the one in Griffin goes to a defendant’s guilt but the one now at issue goes merely to a defendant’s credibility as a witness. See ante, at 69. But it is dominantly in cases where the physical evidence is inconclusive that prosecutors will concentrate all available firepower on the credibility of a testifying defendant. Argument that goes to the defendant’s credibility in such a case also goes to guilt. Indeed, the first sentence of the Court’s account of the trial in this case acknowledges that the questions of guilt and credibility were coextensive. See ante, at 63 (Agard’s trial “ultimately came down to a credibility determination.”).
emphasizes that a prosecutor may make an issue of a defendant’s credibility, and it points for support to our decisions in Jenkins v. Anderson, 447 U. S. 231 (1980), and Brooks v. Tennessee, 406 U. S. 605 (1972). See ante, at 69-70. But again, the distinction between cross-examination and summation is critical. Cross-examination is the criminal trial’s primary means of contesting the credibility of any witness, and a defendant who is also a witness may of course be cross-examined. Jenkins supports the proposition that cross-examination is of sufficient value as an aid to finding truth at trial that prosecutors may sometimes question defendants even about matters that may touch on their constitutional rights, and Brooks suggests that cross-examination can expose a defendant who tailors his testimony. See Jenkins, 447 U. S., at 233, 238; Brooks, 406 U. S., at 609-612. Thus the prosecutor’s tactics in Jenkins and our own counsel in Brooks are entirely consistent with the moderate restriction on prosecutorial license that the Court today rejects.
*88* * *
In the end, we are left with a prosecutorial practice that burdens the constitutional rights of defendants, that cannot be justified by reference to the trial’s aim of sorting guilty defendants from innocent ones, and that is not supported by our case law. The restriction that the Court of Appeals placed on generic accusations of tailoring is both moderate and warranted. That court declared it permissible for the prosecutor to comment on “what the defendant testified to regarding pertinent events” — “the fit between the testimony of the defendant and other witnesses.” 159 F. 8d, at 99. What is impermissible, the Second Circuit held, is simply and only a summation “bolstering ... the prosecution witnesses’ credibility vis-a-vis the defendant’s based solely on the defendant’s exercise of a constitutional right to be present during the trial.” Ibid. I would affirm that sound judgment and therefore dissent from the Court’s disposition.
The prosecutor made the following comment on summation: “A lot of what [the defendant] told you corroborates what the complaining witnesses told you. The only thin[g] that doesn’t is the denials of the crimes. Everything else fits perfectly.” App. 46-47. That, according to the prosecution, is reason for the jury to be suspicious that the defendant falsely tailored his testimony. The implication of this argument seems to be that the more a defendant’s story hangs together, the more likely it is that he is lying. To claim that such an argument helps find truth at trial is to step completely through the looking glass.
The offense charged in Reagan was, moreover, a misdemeanor rather than a felony. See 157 U. S., at 304: Even today, our eases recognize a distinction between serious and petty crimes, and we have held that some provisions of the Sixth Amendment do not apply in petty prosecutions. See, e. g., Lewis v. United States, 518 U. S. 322 (1996) (right to jury trial does not attach in trials for petty offenses). The Reagan Court classified the case before it as belonging to the less serious category of offenses and explicitly denied the defendant the heightened procedural protections that attached in trials for more serious crimes. See 157 U. S., at 302-304.
I do not question the constitutionality of an instruction in which a trial court generally advises the jury that in evaluating the credibility of witnesses, it may take account of the interest of any witness, including the defendant, in the outcome of a case. The interested-witness instruction given in Agard’s case was of this variety. The trial court first told the jury that it should consider the interest that any interested witness might have in the outcome. See Tr. 834 (“If you find that any witness is an interested witness, you should consider such interest in determining the credibility of that person’s testimony and the weight to be given to it.”). It then went on to note, as the Court reports, ante, at 73, that the defendant is an interested witness. See Tr. 834. Any instruction generally applicable to witnesses will affeet defendants who testify, just as the rules governing the admissibility of testimony at trial will restrict defendants’ testimony as they do the testimony of other witnesses. It is a far different matter for an instruction or an argument to impose unique burdens on defendants.
In Brooks v. Tennessee, 406 U. S. 605 (1972), we held this practice unconstitutional under the Fifth and Fourteenth Amendments.
In recent years, several state courts have found it improper for prosecutors to make accusations of tailoring based on the defendant’s constant attendance at trial. See, e. g., State v. Cassidy, 236 Conn. 112, 672 A. 2d 899 (1996); State v. Jones, 580 A. 2d 161, 163 (Me. 1990); Hart v. United States, 538 A. 2d 1146, 1149 (D. C. 1988); State v. Hemingway, 148 Vt. 90, 91-92, 528 A. 2d 746, 747-748 (1987); Commonwealth v. Person, 400 Mass. 136, 138-142, 508 N. E. 2d 88, 90-92 (1987); State v. Johnson, 80 Wash. App. 337, 908 P. 2d 900 (1996). In Commonwealth v. Elberry, 38 Mass. App. 912, 645 N. E. 2d 41 (1995), the trial judge sustained defense counsel’s objection to a prosecutor’s tailoring argument that burdened the defendant’s right to be present at trial and issued the following curative instruction: “Of course, the defendant, who was a witness in this case, was here during the testimony of other witnesses, but he’s got every right to be here, too.... [Y]ou should take everything into consideration in determining credibility, but there is nothing untoward about the defendant being present when other witnesses are testifying.” Id., at 913, 645 N. E. 2d, at 43.
The Court describes the inference now at issue as one not from presence to tailoring but merely from presence to opportunity to tailor. Ante, at 71, n. 2. The proposition that Agard simply had the opportunity to tailor, we note, is not what the prosecutor urged upon the jury. She encouraged the jury to draw, from the fact of Agard’s opportunity, the inference that he had actually tailored his testimony. See App. 49 (Defendant was able “to sit here and listen to the testimony of all the other witnesses before he testifiefd].... [He got] to sit here and think what am I going to say and how am I going to say it? How am I going to fit it into the evidence? . . . He’s a smart man. ... He used everything to his advantage.”)
In fact, the Court of Appeals’ decision in Agard’s case does not tell juries to do anything; it merely prevents prosecutors from inviting them to do something. I presume that the Court means to say that the Court of Appeals’ decision prohibits prosecutors from inviting juries to do something jurors will inevitably do even without invitation. In either case, however, the Court’s confidence that all juries will naturally regard the defendant’s presence at trial as a reason to be suspicious of his testimony is perplexing in light of the Court’s equal confidence that allowing comment on the same subject is “essential” to the truth-finding function of the trial. See ante, at 73. If all juries think this anyway, the pursuit of truth will not suffer if they are not told to think it.