A rule that has minimized the risk of injustice for almost 40 years should not be abandoned without good reason. In 1957, shortly after Congress enacted the statute providing for civil commitment of persons found not guilty by reason of insanity in trials conducted in the District of Columbia, the Court of Appeals, sitting in banc, considered whether juries should be instructed about the significance of that provision. Recognizing that an uninformed jury might erroneously find an insane defendant guilty to avoid the risk that a dangerous individual would otherwise go free, the court held that such an instruction should be given. Lyles v. United States, 254 F. 2d 725 (CADC 1957), cert. denied, 356 U. S. 961 (1958). In an opinion jointly authored by Judge Prettyman and then-Judge Warren Burger, the court explained that the doctrine that the jury has no concern with the consequences of a verdict “does not apply in the problem before us”:
“The issue of insanity having been fairly raised, the jury may return one of three verdicts, guilty, not guilty, or not guilty by reason of insanity. Jurors, in common with people in general, are aware of the meanings of verdicts of guilty and not guilty. . . . But a verdict of not guilty by reason of insanity has no such commonly understood meaning. ... It means neither freedom nor *589punishment. It means the accused will be confined in a hospital for the mentally ill until the superintendent of such hospital certifies, and the court is satisfied, that such person has recovered his sanity and will not in the reasonable future be dangerous to himself or others. We think the jury has a right to know the meaning of this possible verdict as accurately as it knows by common knowledge the meaning of the other two possible verdicts.” Lyles, 254 F. 2d, at 728.
Concurring with this part of the foregoing opinion, Judge Bazelon acknowledged that “[t]he false assumption that acquittal by reason of insanity, like outright acquittal, frees the accused to walk out on the streets may lead juries to convict, despite strong evidence of insanity at the time of the crime.” Id., at 734. Trial courts in the District of Columbia have used a pattern instruction — approved by prosecutors, defense counsel, and trial judges — ever since.1
Other federal courts did not give a comparable instruction prior to 1984 because no federal statute authorized civil commitment for insanity acquittees except in the District of Columbia. In those courts, an instruction advising the jury about the consequences of a verdict of not guilty by reason of insanity — often that such a defendant would, indeed, go free — would have tended to increase the risk of improper convictions. It was therefore appropriate for federal judges to adhere to the general rule that the jury should be instructed to base its decision on the evidence before it, with*590out regard to the possible consequences of its verdict. That rule, of course, was primarily designed to protect defendants from the risk that jurors might otherwise improperly rely on matters such as sympathy for the victim, arguments of counsel, or inadmissible comments in the courtroom.
When Congress enacted the Insanity Defense Reform Act of 1984 (Act), 18 U.S.C. §§17, 4241-4247, it established a civil commitment process for the entire federal system, thus making the basis for the District of Columbia Circuit’s holding in Lyles applicable to all federal courts. The Act’s legislative history unmistakably demonstrates that the Act’s sponsors assumed that the Lyles precedent would thereafter be followed nationwide. See ante, at 583. That assumption does not have the force of a statutory mandate, but it verifies that thoughtful legislators familiar with the issue believed that precedent to be entirely sound. That this Court should now decide to change an established rule that Congress accepted and that protects defendants meaningfully against an obvious risk of injustice is startling — particularly when that change is for no reason other than a perceived inconsistency with another rule that is generally protective of defendants’ rights. A far wiser disposition would allow the defendant to choose between the two rules, rather than tilt the scales to favor the prosecutor in every case.
The incongruity of the Court’s holding is compounded by its selection of Rogers v. United States, 422 U. S. 35 (1975), as its authority for what it calls the “principle” that juries should not consider the consequences of their verdict. Ante, at 579. It is worth noting that the writer of the Court’s opinion in Rogers — Chief Justice Burger — was also one of the authors of Lyles. In Rogers, the jury had sent the judge a note asking whether he would accept a verdict of “Guilty as charged with extreme mercy of the Court”; when the court answered yes, the jury returned five minutes later with that verdict. Rogers, 422 U. S., at 36-37. What Rogers held is *591that the guilty verdict had to be set aside because the court had violated Rule 43 of the Federal Rules of Criminal Procedure by responding to an inquiry from the jury without advising defense counsel. Id., at 40-41. The Court also considered the judge’s response to be misleading because it did not advise the jury that their recommendation of mercy would not be binding on the court. Ibid. In that context, the failure to admonish the jury that it should reach its verdict without regard to what sentence might be imposed was prejudicial to the defendant. Instead of supporting the majority’s view, the case is more relevant for its illustration of how concerned juries are about the actual consequences of their verdicts. When there is a realistic danger that jurors’ deliberations may be distorted by an incorrect assumption about those consequences, elementary notions of fairness demand that a clarifying instruction be given.
The Court suggests that the instruction might actually prejudice the defendant. Ante, at 585-586. That argument lacks merit, as there is no need to give the instruction unless the defendant requests it. Alternatively, the Court advances the tired argument that if we followed the practice approved in Lyles, “the rule against informing jurors of the consequences of their verdicts would soon be swallowed by the exceptions,” ante, at 587. Given that the Lyles rule has survived in the District since 1957 without such consequences, this concern is illusory. Some courts have assumed that the instruction would help jurors focus on issues of guilt instead of punishment. “Freed from confusion and fear as to the practical effect of a verdict of not guilty by reason of insanity, jurors should be able to decide the insanity issue solely on the evidence and law governing the defense.” State v. Shickles, 760 P. 2d 291, 298 (Utah 1988). Rather than relying on a totally unsubstantiated qualm belied by history, it would be far wiser for the Court simply to recognize both the seriousness of the harm that may result from *592the refusal to give the instruction and the absence of any identifiable countervailing harm that may result from giving it.
The Court also contends that jurors today are more familiar with the consequences of a verdict of not guilty by reason of insanity than they were in 1957 when Lyles was decided. Ante, at 584, n. 9. No one has suggested, however, that the level of understanding even approximates that of the conventional choice between “guilty” and “not guilty.” Indeed, one recent study concluded that “the public overestimates the extent to which insanity acquittees are released upon acquittal and underestimates the extent to which they are hospitalized as well as the length of confinement of insanity acquittees who are sent to mental hospitals.”2 As long as significant numbers of potential jurors believe that an insanity acquittee will be released at once, the instruction serves a critical purpose. Yet even if, as the Court seems prepared to assume, all jurors are already knowledgeable about the issue, surely telling them what they already know can do no harm.
An increasing number of States that have considered the question endorses use of the instruction,3 as has the American Bar Association.4 Judge Newman’s succinct assessment *593of the pros and cons is exactly right: “There is no reason to keep this information from the jurors and every reason to make them aware of it.” United States v. Blume, 967 F. 2d 45, 52 (CA2 1992) (concurring opinion).
I respectfully dissent.
Instruction 5.11 in the 1978 edition of the District of Columbia Criminal Jury Instructions reads:
“If the defendant is found not guilty by reason of insanity it becomes the duty of the court to commit him to St. Elizabeths Hospital. There will be a hearing within 50 days to determine whether the defendant is entitled to release. In that hearing the defendant has the burden of proof. The defendant will remain in custody, and will be entitled to release from custody only if the court finds by a preponderance of the evidence that he is not likely to injure himself or other persons due to mental illness.”
Silver, Cirincione, and Steadman, Demythologizing Inaccurate Perceptions of the Insanity Defense, 18 Law and Human Behavior 63, 68 (Feb. 1994).
See, e. g., Erdman v. State, 315 Md. 46, 553 A. 2d 244 (1989); State v. Shickles, 760 P. 2d 291 (Utah 1988); People v. Young, 189 Cal. App. 3d 891, 234 Cal. Rptr. 819 (1987); People v. Thomson, 197 Colo. 232, 591 P. 2d 1031 (1979); Commonwealth v. Mulgrew, 475 Pa. 271, 380 A. 2d 349 (1977); Roberts v. State, 335 So. 2d 285 (Fla. 1976); Commonwealth v. Mutina, 366 Mass. 810, 323 N. E. 2d 294 (1975); State v. Babin, 319 So. 2d 367 (La. 1975). See also Fleming, Instructions in State Criminal Case in Which Defendant Pleads Insanity as to Hospital Confinement in Event of Acquittal, 81 A. L. R. 4th 659, 667 (1990) (noting “an apparent trend toward requiring or authorizing a jury instruction on the legal consequences of an insanity acquittal”).
ABA Criminal Justice Mental Health Standards § 7-6.8 (1989).