announced the judgment of the Court and delivered the opinion of the Court with respect to Parts I, II, and IV, and an opinion with respect to Parts III and V, in which Justice Kennedy joins.
In both of these cases, respondents were tried for criminal contempt of court for violating court orders that prohibited them from engaging in conduct that was later the subject of a criminal prosecution. We consider whether the subsequent criminal prosecutions are barred by the Double Jeopardy Clause.
I
Respondent Alvin Dixon was arrested for second-degree murder and was released on bond. Consistent with the District of Columbia’s bail law authorizing the judicial officer to impose any condition that “will reasonably assure the appearance of the person for trial or the safety of any other person or the community,” D. C. Code Ann. §23-1321(a) (1989), Dixon’s release form specified that he was not to commit “any criminal offense,” and warned that any violation of the conditions of release would subject him “to revocation of release, an order of detention, and prosecution for contempt of court.” See D. C. Code Ann. § 23-1329(a) (1989) (authorizing those sanctions).
While awaiting trial, Dixon was arrested and indicted for possession of cocaine with intent to distribute, in violation of D. C. Code Ann. § 33-541(a)(l) (1988). The court issued an order requiring Dixon to show cause why he should not be held in contempt or have the terms of his pretrial release modified. At the show-cause hearing, four police officers testified to facts surrounding the alleged drug offense; Dixon’s counsel cross-examined these witnesses and introduced other evidence. The court concluded that the Government had established “‘beyond a reasonable doubt that [Dixon] was in possession of drugs and that those drugs were possessed with the intent to distribute.’” 598 A. 2d 724, 728 (D. C. 1991). The court therefore found Dixon guilty of *694criminal contempt under § 23-1329(c), which allows contempt sanctions after expedited proceedings without a jury and “in accordance with principles applicable to proceedings for criminal contempt.” For his contempt, Dixon was sentenced to 180 days in jail. § 23 — 1329(c) (maximum penalty of six months’ imprisonment and $1,000 fine). He later moved to dismiss the cocaine indictment on double jeopardy grounds; the trial court granted the motion.
Respondent Michael Foster’s route to this Court is similar. Based on Foster’s alleged physical attacks upon her in the past, Foster’s estranged wife Ana obtained a civil protection order (CPO) in Superior Court of the District of Columbia. See D. C. Code Ann. § 16-1005(c) (1989) (CPO may be issued upon a showing of good cause to believe that the subject “has committed or is threatening an intrafamily offense”). The order, to which Foster consented, required that he not “‘molest, assault, or in any manner threaten or physically abuse’” Ana Foster; a separate order, not implicated here, sought to protect her mother. 598 A. 2d, at 725-726.
Over the course of eight months, Ana Foster filed three separate motions to have her husband held in contempt for numerous violations of the CPO. Of the 16 alleged episodes, the only charges relevant here are three separate instances of threats (on November 12,1987, and March 26 and May 17, 1988) and two assaults (on November 6, 1987, and May 21, 1988), in the most serious of which Foster “threw [his wife] down basement stairs, kicking her body[,] . . . pushed her head into the floor causing head injuries, [and Ana Foster] lost consciousness.” 598 A. 2d, at 726.
After issuing a notice of hearing and ordering Foster to appear, the court held a 3-day bench trial. Counsel for Ana Foster and her mother prosecuted the action; the United States was not represented at trial, although the United States Attorney was apparently aware of the action, as was the court aware of a separate grand jury proceeding on some of the alleged criminal conduct. As to the assault charges, *695the court stated that Ana Foster would have “to prove as an element, first that there was a Civil Protection Order, and then [that]... the assault as defined by the criminal code, in fact occurred.” Tr. in Nos. IF-630-87, IF-631-87 (Aug. 8, 1988), p. 367; accord, id., at 368. At the close of the plaintiffs’ case, the court granted Foster’s motion for acquittal on various counts, including the alleged threats on November 12 and May 17. Foster then took the stand and generally denied the allegations. The court found Foster guilty beyond a reasonable doubt of four counts of criminal contempt (three violations of Ana Foster’s CPO, and one violation of the CPO obtained by her mother), including the November 6, 1987, and May 21, 1988, assaults, but acquitted him on other counts, including the March 26 alleged threats. He was sentenced to an aggregate 600 days’ imprisonment. See § 16 — 1005(f) (authorizing contempt punishment); Super. Ct. of D. C. Intrafamily Rules 7(c), 12(e) (1987) (maximum punishment of six months’ imprisonment and $300 fine).
The United States Attorney’s Office later obtained an indictment charging Foster with simple assault on or about November 6, 1987 (Count I, violation of § 22-504); threatening to injure another on or about November 12, 1987, and March 26 and May 17, 1988 (Counts II-IV, violation of § 22-2307); and assault with intent to kill on or about May 21, 1988 (Count V, violation of §22-501). App. 43-44. Ana Foster was the complainant in all counts; the first and last counts were based on the events for which Foster had been held in contempt, and the other three were based on the alleged events for which Foster was acquitted of contempt. Like Dixon, Foster filed a motion to dismiss, claiming a double jeopardy bar to all counts, and also collateral estoppel as to Counts II-IV. The trial court denied the double jeopardy claim and did not rule on the collateral-estoppel assertion.
The Government appealed the double jeopardy ruling in Dixon, and Foster appealed the trial court’s denial of his motion. The District of Columbia Court of Appeals consoli*696dated the two cases, reheard them en banc, and, relying on our recent decision in Grady v. Corbin, 495 U. S. 508 (1990), ruled that both subsequent prosecutions were barred by the Double Jeopardy Clause. 598 A. 2d, at 725. In its petition for certiorari, the Government presented the sole question “[wjhether the Double Jeopardy Clause bars prosecution of a defendant on substantive criminal charges based upon the same conduct for which he previously has been held in criminal contempt of court.” Pet. for Cert. I. We granted certiorari, 503 U. S. 1004 (1992).
II
To place these cases in context, one must understand that they are the consequence of a historically anomalous use of the contempt power. In both Dixon and Foster, a court issued an order directing a particular individual not to commit criminal offenses. (In Dixon’s case, the court incorporated the entire criminal code; in Foster’s case, the criminal offense of simple assault.) That could not have occurred at common law, or in the 19th-century American judicial system.
At common law, the criminal contempt power was confined to sanctions for conduct that interfered with the orderly administration of judicial proceedings. 4 W. Blackstone, Commentaries *280-*285. That limitation was closely followed in American courts. See United States v. Hudson, 7 Cranch 32, 34 (1812); R. Goldfarb, The Contempt Power 12-20 (1963). Federal courts had power to “inforce the observance of order,” but those “implied powers” could not support common-law jurisdiction over criminal acts. Hudson, supra, at 34. In 1831, Congress amended the Judiciary Act of 1789, allowing federal courts the summary contempt power to punish generally “disobedience or resistance” to court orders. §1, Act of March 2, 1831, 4 Stat. 487-488. See Bloom v. Illinois, 391 U. S. 194, 202-204 (1968) (discussing evolution of federal courts’ statutory contempt power).
*697The 1831 amendment of the Judiciary Act still would not have given rise to orders of the sort at issue here, however, since there was a long common-law tradition against judicial orders prohibiting violation of the law. Injunctions, for example, would not issue to forbid infringement of criminal or civil laws, in the absence of some separate injury to private interest. See, e.g., 3 Blackstone, supra, at *426, n. 1; J. High, Law of Injunctions §23, pp. 15-17, and notes (1873) (citing English cases); C. Beach, Law of Injunctions §§ 58-59, pp. 71-73 (1895) (same). The interest protected by the criminal or civil prohibition was to be vindicated at law — and though equity might enjoin harmful acts that happened to violate civil or criminal law, it would not enjoin violation of civil or criminal law as such. See, e. g., Sparhawk v. Union Passenger R. Co., 54 Pa. St. 401, 422-424 (1867) (refusing to enjoin railroad’s violation of Sunday closing law); Attorney General v. Utica Insurance Co., 2 Johns. Ch. 371, 378 (N. Y. 1817) (refusing to enjoin violation of banking statute).
It is not surprising, therefore, that the double jeopardy issue presented here — whether prosecution for criminal contempt based on violation of a criminal law incorporated into a court order bars a subsequent prosecution for the criminal offense — did not arise at common law, or even until quite recently in American cases. See generally Zitter, Contempt Finding as Precluding Substantive Criminal Charges Relating to Same Transaction, 26 A. L. R. 4th 950, 953-956 (1983). English and earlier American cases do report instances in which prosecution for criminal contempt of court — as originally understood — did not bar a subsequent prosecution for a criminal offense based on the same conduct. See, e. g., King v. Lord Ossulston, 2 Str. 1107, 93 Eng. Rep. 1063 (K. B. 1739); State v. Yancy, 4 N. C. 133 (1814). But those contempt prosecutions were for disruption of judicial process, in which the disruptive conduct happened also to be criminal.
The Double Jeopardy Clause, whose application to this new context we are called upon to consider, provides that no *698person shall “be subject for the same offence to be twice put in jeopardy of life or limb.” U. S. Const., Amdt. 5. This protection applies both to successive punishments and to successive prosecutions for the same criminal offense. See North Carolina v. Pearce, 395 U. S. 711 (1969). It is well established that criminal contempt, at least the sort enforced through nonsummary proceedings, is “a crime in the ordinary sense.” Bloom, supra, at 201. Accord, New Orleans v. Steamship Co., 20 Wall. 387, 392 (1874).
We have held that constitutional protections for criminal defendants other than the double jeopardy provision apply in nonsummary criminal contempt prosecutions just as they do in other criminal prosecutions. See, e. g., Gompers v. Bucks Stove & Range Co., 221 U. S. 418, 444 (1911) (presumption of innocence, proof beyond a reasonable doubt, and guarantee against self-incrimination); Cooke v. United States, 267 U. S. 517, 537 (1925) (notice of charges, assistance of counsel, and right to present a defense); In re Oliver, 333 U. S. 257, 278 (1948) (public trial). We think it obvious, and today hold, that the protection of the Double Jeopardy Clause likewise attaches. Accord, Menna v. New York, 423 U. S. 61 (1975) (per curiam); Colombo v. New York, 405 U. S. 9 (1972) (per curiam).
In both the multiple punishment and multiple prosecution contexts, this Court has concluded that where the two offenses for which the defendant is punished or tried cannot survive the “same-elements” test, the double jeopardy bar applies. See, e. g., Brown v. Ohio, 432 U. S. 161, 168-169 (1977); Blockburger v. United States, 284 U. S. 299, 304 (1932) (multiple punishment); Gavieres v. United States, 220 U. S. 338, 342 (1911) (successive prosecutions). The same-elements test, sometimes referred to as the “Blockburger” test, inquires whether each offense contains an element not contained in the other; if not, they are the “same offence” and double jeopardy bars additional punishment and successive prosecution. In a case such as Yancy, for example, in which *699the contempt prosecution was for disruption of judicial business, the same-elements test would not bar subsequent prosecution for the criminal assault that was part of the disruption, because the contempt offense did not require the element of criminal conduct, and the criminal offense did not require the element of disrupting judicial business.1
We recently held in Grady that in addition to passing the Blockburger test, a subsequent prosecution must satisfy a “same-conduct” test to avoid the double jeopardy bar. The Grady test provides that, “if, to establish an essential element of an offense charged in that prosecution, the government will prove conduct that constitutes an offense for which the defendant has already been prosecuted,” a second prosecution may not be had. 495 U. S., at 510.
Ill
A
The first question before us today is whether Blockburger analysis permits subsequent prosecution in this new criminal contempt context, where judicial order has prohibited criminal act. If it does, we must then proceed to consider whether Grady also permits it. See Grady, supra, at 516.
We begin with Dixon. The statute applicable in Dixon’s contempt prosecution provides that “[a] person who has been conditionally released . . . and who has violated a condition of release shall be subject to ... prosecution for contempt of court.” §23-1329(a). Obviously, Dixon could not commit an “offence” under this provision until an order setting out conditions was issued. The statute by itself imposes no legal obligation on anyone. Dixon’s cocaine possession, although an offense under D. C. Code Ann. § 33-541(a) (1988 and Supp. 1992), was not an offense under § 23-1329 until a *700judge incorporated the statutory drug offense into his release order.
In this situation, in which the contempt sanction is imposed for violating the order through commission of the incorporated drug offense, the later attempt to prosecute Dixon for the drug offense resembles the situation that produced our judgment of double jeopardy in Harris v. Oklahoma, 433 U. S. 682 (1977) (per curiam). There we held that a subsequent prosecution for robbery with a firearm was barred by the Double Jeopardy Clause, because the defendant had already been tried for felony murder based on the same underlying felony. We have described our terse per curiam in Harris as standing for the proposition that, for double jeopardy purposes, “the crime generally described as felony murder” is not “a separate offense distinct from its various elements.” Illinois v. Vitale, 447 U. S. 410, 420-421 (1980). Accord, Whalen v. United States, 445 U. S. 684, 694 (1980). So too here, the “crime” of violating a condition of release cannot be abstracted from the “element” of the violated condition. The Dixon court order incorporated the entire governing criminal code in the same manner as the Harris felony-murder statute incorporated the several enumerated felonies. Here, as in Harris, the underlying substantive criminal offense is “a species of lesser-included offense.”2 Vitale, supra, at 420. Accord, Whalen, supra.
*701To oppose this analysis, the Government can point only to dictum in In re Debs, 158 U. S. 564, 594, 599-600 (1895), which, to the extent it attempted to exclude certain nonsummary contempt prosecutions from various constitutional protections for criminal defendants, has been squarely rejected by cases such as Bloom, 391 U. S., at 208. The Government also relies upon In re Chapman, 166 U. S. 661 (1897), and Jurney v. MacCracken, 294 U. S. 125 (1935), which recognize Congress’ power to punish as contempt the refusal of a witness to testify before it. But to say that Congress can punish such a refusal is not to say that a criminal court can punish the same refusal yet again. Neither case dealt with that issue, and Chapman specifically declined to address it, noting that successive prosecutions (before Congress for contemptuous refusal to testify and before a court for violation of a federal statute making such refusal a crime) were “improbable.” 166 U. S., at 672.
Both the Government, Brief for United States 15-17, and Justice Blackmun, post, at 743, contend that the legal obligation in Dixon’s case may serve “interests . . . fundamentally different” from the substantive criminal law, because it derives in part from the determination of a court rather than a determination of the legislature. That distinction seems questionable, since the court’s power to establish conditions of release, and to punish their violation, was conferred by statute; the legislature was the ultimate source of both the criminal and the contempt prohibition. More importantly, however, the distinction is of no moment for purposes of the Double Jeopardy Clause, the text of which looks to whether the offenses are the same, not the interests that the offenses violate. And this Court stated long ago that criminal con*702tempt, at least in its nonsummary form, “is a crime in every fundamental respect.” Bloom, supra, at 201; accord, e. g., Steamship Co., 20 Wall., at 392. Because Dixon’s drug offense did not include any element not contained in his previous contempt offense, his subsequent prosecution violates the Double Jeopardy Clause.
The foregoing analysis obviously applies as well to Count I of the indictment against Foster, charging assault in violation of § 22-504, based on the same event that was the subject of his prior contempt conviction for violating the provision of the CPO forbidding him to commit simple assault under §22-504.3 The subsequent prosecution for assault fails the Blockburger test, and is barred.4
B
The remaining four counts in Foster, assault with intent to kill (Count V; §22-501) and threats to injure or kidnap (Counts II-IV; §22-2307), are not barred under Blockburger. As to Count V: Foster’s conduct on May 21, 1988, was found to violate the Family Division’s order that he not “molest, assault, or in any manner threaten or physically abuse” his wife. At the contempt hearing, the court stated that Ana *703Foster’s attorney, who prosecuted the contempt, would have to prove, first, knowledge of a CPO, and, second, a willful violation of one of its conditions, here simple assault as defined by the criminal code.5 See, e. g., 598 A. 2d, at 727-728; In re Thompson, 454 A. 2d 1324, 1326 (D. C. 1982); accord, Parker v. United States, 373 A. 2d 906, 907 (D. C. 1977) (per curiam). On the basis of the same episode, Foster was then indicted for violation of §22-501, which proscribes assault with intent to kill. Under governing law, that offense requires proof of specific intent to kill; simple assault does not.6 See Logan v. United States, 483 A. 2d 664, 672-673 (D. C. 1984). Similarly, the contempt offense required proof of knowledge of the CPO, which assault with intent to kill does not. Applying the Blockburger elements test, the result is clear: These crimes were different offenses, and the sub*704sequent prosecution did not violate the Double Jeopardy Clause.7
Counts II, III, and IV of Foster’s indictment are likewise not barred. These charged Foster under §22-2307 (forbidding anyone to “threate[n] ... to kidnap any person or to injure the person of another or physically damage the property of any person”) for his alleged threats on three separate dates. Foster’s contempt prosecution included charges that, on the same dates, he violated the CPO provision ordering that he not “in any manner threaten” Ana Foster. Conviction of the contempt required willful violation of the CPO— which conviction under §22-2307 did not; and conviction under § 22-2307 required that the threat be a threat to kidnap, to inflict bodily injury, or to damage property — which conviction of the contempt (for violating the CPO provision that Foster not “in any manner threaten”) did not.8 Each *705offense therefore contained a separate element, and the Blockburger test for double jeopardy was not met.
IV
Having found that at least some of the counts at issue here are not barred by the Blockburger test, we must consider whether they are barred by the new, additional double jeopardy test we announced three Terms ago in Grady v. Corbin.9 They undoubtedly are, since Grady prohibits “a subsequent prosecution if, to establish an essential element of an offense charged in that prosecution [here, assault as an element of assault with intent to kill, or threatening as an element of threatening bodily injury], the government will *706prove conduct that constitutes an offense for which the defendant has already been prosecuted [here, the assault and the threatening, which conduct constituted the offense of violating the CPO].” 495 U. S., at 510.
We have concluded, however, that Grady must be overruled. Unlike Blockburger analysis, whose definition of what prevents two crimes from being the “same offence,” U. S. Const., Arndt. 5, has deep historical roots and has been accepted in numerous precedents of this Court, Grady lacks constitutional roots. The “same-conduct” rule it announced is wholly inconsistent with earlier Supreme Court precedent and with the clear common-law understanding of double jeopardy. See, e. g., Gavieres v. United States, 220 U. S., at 345 (in subsequent prosecution, “[wjhile it is true that the conduct of the accused was one and the same, two offenses resulted, each of which had an element not embraced in the other”). We need not discuss the many proofs of these statements, which were set forth at length in the Grady dissent. See 495 U. S., at 526 (opinion of SCALIA, J.). We will respond, however, to the contrary contentions of today’s pro-Grady dissents.
The centerpiece of Justice Souter’s analysis is an appealing theory of a “successive prosecution” strand of the Double Jeopardy Clause that has a different meaning from its supposed “successive punishment” strand. We have often noted that the Clause serves the function of preventing both successive punishment and successive prosecution, see, e. g., North Carolina v. Pearce, 395 U. S. 711 (1969), but there is no authority, except Grady, for the proposition that it has different meanings in the two contexts. That is perhaps because it is embarrassing to assert that the single term “same offence” (the words of the Fifth Amendment at issue here) has two different meanings — that what is the same offense is yet not the same offense. Justice Souter provides no authority whatsoever (and we are aware of none) for the bald assertion that “we have long held that [the government] *707must sometimes bring its prosecutions for [separate] offenses together.” Post, at 747. The collateral-estoppel effect attributed to the Double Jeopardy Clause, see Ashe v. Swenson, 397 U. S. 436 (1970), may bar a later prosecution for a separate offense where the Government has lost an earlier prosecution involving the same facts. But this does not establish that the Government “must... bring its prosecutions . . . together.” It is entirely free to bring them separately, and can win convictions in both. Of course the collateralestoppel issue is not raised in this case.
Justice Souter relies upon four cases to establish the existence of some minimal antecedents to Grady. Post, at 749-758. The fountainhead of the “same-conduct” rule, he asserts, is In re Nielsen, 131 U. S. 176 (1889). That is demonstrably wrong. Nielsen simply applies the common proposition, entirely in accord with Blockburger, that prosecution for a greater offense (cohabitation, defined to require proof of adultery) bars prosecution for a lesser included offense (adultery). That is clear from the Nielsen Court’s framing of the question (“Being of opinion, therefore, that habeas corpus was a proper remedy for the petitioner, if the crime of adultery with which he was charged was included in the crime of unlawful cohabitation for which he was convicted and punished, that question is now to be considered,” 131 U. S., at 185 (emphasis added)), from its legal analysis, id., at 186-189, and from its repeated observations that cohabitation required proof of adultery, id., at 187, 189.10
*708His second case comes almost a century later. Brown v. Ohio, 432 U. S. 161 (1977), contains no support for his position except a footnote that cites Nielsen for the proposition that “[t]he Blockburger test is not the only standard for determining whether successive prosecutions impermissibly involve the same offense.” Brown, supra, at 166-167, n. 6. Not only is this footnote the purest dictum, but it flatly contradicts the text of the opinion which, on the very next page, describes Nielsen as the first Supreme Court case to endorse the Blockburger rule. Brown, supra, at 168. Quoting that suspect dictum multiple times, see post, at 748, 754, cannot convert it into case law. See United States Nat. Bank of Ore. v. Independent Ins. Agents of America, Inc., 508 U. S. 439, 463, n. 11 (1993) (emphasizing “the need to distinguish an opinion’s holding from its dicta”). The holding of Brown, like that of Nielsen, rests squarely upon the existence of a lesser included offense. 432 U. S., at 162 (setting out question presented).
The third case is Harris, which Justice Souter asserts was a reaffirmation of what he contends was the earlier holding in Nielsen, that the Blockburger test is “insufficien[t] for determining when a successive prosecution [is] barred,” and that conduct, and not merely elements of the offense, must be the object of inquiry. Post, at 755. Surely not. Harris never uses the word “conduct,” and its entire discussion focuses on the elements of the two offenses. See, e. g., 433 U. S., at 682-683, n. (to prove felony murder, “it was necessary for all the ingredients of the underlying felony” to be proved). Far from validating Justice Souter’s extraordinarily implausible reading of Nielsen, Harris plainly rejects that reading, treating the earlier case as having focused (like Blockburger) upon the elements of the offense. Immedi*709ately after stating that conviction for felony murder, a “greater crime,” “cannot be had without conviction of the lesser crime,” the Harris Court quotes Nielsen’s statement that “‘a person [who] has been tried and convicted for a crime which has various incidents included in it, . . . cannot be a second time tried for one of those incidents.’” 433 U. S., at 682-683, quoting from 131 U. S., at 188. It is clear from that context that Harris regarded “incidents included” to mean “offenses included” — a reference to defined crimes rather than to conduct.
Finally, Justice Souter misdescribes Vitale. Despite his bold assertion to the contrary, see post, at 757, Vitale unquestionably reads Harris as merely an application of the double jeopardy bar to lesser and greater included offenses.11 Justice Souter instead elevates the statement in Vitale that, on certain hypothetical facts, the petitioner would have a “substantial” “claim” of double jeopardy on a Grady-type theory, see post, at 756-757, into a holding that the petitioner would win on that theory. Post, at 757, 763. No Justice, the Vitale dissenters included, has ever construed this passage as answering, rather than simply raising, the question on which we later granted certiorari in Grady. See 447 U. S., at 426 (Stevens, J., dissenting) (in addition to finding the same-conduct claim “substantial,” dissent would find it “dispositive”). See also Grady, 495 U. S., at 510 (Vitale “suggested” same-conduct test adopted in Grady).
In contrast to the above-discussed dicta relied upon by Justice Souter, there are two pre-Grady (and post-Nielsen) cases that are directly on point. In both Gavieres v. United States, 220 U. S., at 343, and Burton v. United States, 202 U. S. 344, 379-381 (1906), the Court upheld subse*710quent prosecutions after concluding that the Blockburger test (and only the Blockburger test) was satisfied.12 These cases are incompatible with the belief that Nielsen had created an additional requirement beyond the “elements” standard.13 Totally ignored by Justice Souter are the *711many early American cases construing the Double Jeopardy Clause, which support only an “elements” test. See Grady, supra, at 533-535 (Scalia, J., dissenting).14
But Grady was not only wrong in principle; it has already proved unstable in application. Less than two years after it came down, in United States v. Felix, 503 U. S. 378 (1992), we were forced to recognize a large exception to it. There we concluded that a subsequent prosecution for conspiracy to manufacture, possess, and distribute methamphetamine was not barred by a previous conviction for attempt to manufacture the same substance. We offered as a justification for avoiding a “literal” (i. e., faithful) reading of Grady “longstanding authority” to the effect that prosecution for conspiracy is not precluded by prior prosecution for the substantive offense. Felix, supra, at 388-391. Of course the very existence of such a large and longstanding “exception” to the *712Grady rule gave cause for concern that the rule was not an accurate expression of the law. This “past practice” excuse is not available to support the ignoring of Grady in the present case, since there is no Supreme Court precedent even discussing this fairly new breed of successive prosecution (criminal contempt for violation of a court order prohibiting a crime, followed by prosecution for the crime itself).
A hypothetical based on the facts in Harris reinforces the conclusion that Grady is a continuing source of confusion and must be overruled. Suppose the State first tries the defendant for felony murder, based on robbery, and then indicts the defendant for robbery with a firearm in the same incident. Absent Grady, our cases provide a clear answer to the double jeopardy claim in this situation. Under Blockburger, the second prosecution is not barred — as it clearly was not barred at common law, as a famous case establishes. In King v. Vandercomb, 2 Leach. 708, 717, 168 Eng. Rep. 455, 460 (K. B. 1796), the government abandoned, midtrial, prosecution of defendant for burglary by breaking and entering and stealing goods, because it turned out that no property had been removed on the date of the alleged burglary. The defendant was then prosecuted for burglary by breaking and entering with intent to steal. That second prosecution was allowed, because “these two offences are so distinct in their nature, that evidence of one of them will not support an indictment for the other.” Ibid. Accord, English and American cases cited in Grady, 495 U. S., at 532-535 (Scalia, J., dissenting).15
*713Having encountered today yet another situation in which the pre-Grady understanding of the Double Jeopardy Clause allows a second trial, though the “same-conduct” test would not, we think it time to acknowledge what is now, three years after Grady, compellingly clear: The case was a mistake. We do not lightly reconsider a precedent, but, because Grady contradicted an “unbroken line of decisions,” contained “less than accurate” historical analysis, and has produced “confusion,”16 we do so here. Solorio v. United States, 483 U. S. *714435, 439, 442, 450 (1987). Although stare decisis is the “preferred course” in constitutional adjudication, “when governing decisions are unworkable or are badly reasoned, ‘this Court has never felt constrained to follow precedent.’” Payne v. Tennessee, 501 U. S. 808, 827 (1991) (quoting Smith v. Allwright, 321 U. S. 649, 665 (1944), and collecting examples). We would mock stare decisis and only add chaos to our double jeopardy jurisprudence by pretending that Grady survives when it does not. We therefore accept the Government’s invitation to overrule Grady, and Counts II, III, IV, and V of Foster’s subsequent prosecution are not barred.17
V
Dixon’s subsequent prosecution, as well as Count I of Foster’s subsequent prosecution, violate the Double Jeopardy Clause.18 For the reasons set forth in Part IV, the other counts of Foster’s subsequent prosecution do not violate the Double Jeopardy Clause.19 The judgment of the District of Columbia Court of Appeals is affirmed in part and reversed in part, and the case is remanded for proceedings not inconsistent with this opinion.
It is so ordered.
State v. Yancy, 4 N. C. 133 (1814), it should be noted, involved what is today called summary contempt. We have not held, and do not mean by this example to decide, that the double jeopardy guarantee applies to such proceedings.
In order for the same analysis to be applicable to violation of a statute criminalizing disobedience of a lawful police order, as The Chief Justice’s dissent on this point hypothesizes, see post, at 719, the statute must embrace police “orders” that “command” the noncommission of crimes— for instance, “Don’t shoot that man!” It seems to us unlikely that a “police order” statute would be interpreted in this fashion, rather than as addressing new obligations imposed by lawful order of police (for example, the obligation to remain behind police lines, or to heed a command to “Freeze!”). If, however, such a statute were interpreted to cover police orders forbidding crimes, the Double Jeopardy Clause would as a practical matter bar subsequent prosecution only for relatively minor offenses, such *701as assault (the only conceivable lesser included offense of an order not to “shoot”) — unless one assumes that constables often order the noncommission of serious crimes (for example, “Don’t murder that man!”) and that serious felons such as murderers are first prosecuted for disobeying police orders.
It is not obvious that the word “assault” in the CPO bore the precise meaning “assault under § 22-504.” The court imposing the contempt construed it that way, however, and the point has not been contested in this litigation.
Justice White complains that this section of our opinion gives the arguments of the United States “short shrift,” post, at 720, and treats them in “conclusory” fashion, post, at 721. He then proceeds to reject these arguments, largely by agreeing with our analysis, post, at 721, 722, 724, 726. We think it unnecessary, and indeed undesirable, to address at any greater length than we have arguments based on dictum and inapplicable doctrines such as dual sovereignty. The remainder of that part of Justice White’s opinion that deals with this issue argues — by no means in conclusory fashion — that its practical consequences for law enforcement are not serious. Post, at 727-731. He may be right. But we do not share his “pragmatic” view, post, at 739, that the meaning of the Double Jeopardy Clause depends upon our approval of its consequences.
Given this requirement of willful violation of the order, Justice White’s desire to “put to the side the CPO,” because it only “triggered the court’s authority” cannot be reconciled with his desire to “comparte] the substantive offenses of which respondents stood accused.” Post, at 734. The “substantive offense” of criminal contempt is willful violation of a court order. Far from a mere jurisdictional device, that order (or CPO) is the centerpiece of the entire proceeding. Its terms define the prohibited conduct, its existence supports imposition of a criminal penalty, and willful violation of it is necessary for conviction. To ignore the CPO when determining whether two offenses are the “same” is no more possible than putting aside the statutory definitions of criminal offenses. Of course, Justice White’s view that the elements of criminal contempt are essentially irrelevant for double jeopardy analysis does have precedent— albeit erroneous — in Grady’s same-conduct test. Grady v. Corbin, 495 U. S. 508 (1990). Justice Souter also ignores the knowledge element. Post, at 761, n. 10.
We accept, as we ordinarily do, the construction of a District of Columbia law adopted by the District of Columbia Court of Appeals. See, e. g., Pernell v. Southall Realty, 416 U. S. 363, 368-369 (1974). The construction here has sound support in the text of the statute. Compare D. C. Code Ann. § 22-501 (1989) (assault with intent to kill, rob, rape, or poison) with § 22-504 (assault).
Justice White’s suggestion, post, at 737-738, that if Foster received a lesser-included-offense instruction on assault at his trial for assault with intent to kill, we would uphold a conviction on that lesser count is simply wrong. Under basic Blockburger analysis, Foster may neither be tried a second time for assault nor again convicted for assault, as we have concluded as to Count I (charging simple assault). Thus, Foster certainly does receive the “full constitutional protection to which he is entitled,” post, at 738, n. 10: he may neither be tried nor convicted a second time for assault. That does not affect the conclusion that trial and conviction for assault with intent to kill are not barred. It merely illustrates the unremarkable fact that one offense (simple assault) may be an included offense of two offenses (violation of the CPO for assault, and assault with intent to kill) that are separate offenses under Blockburger.
We think it is highly artificial to interpret the CPO’s prohibition of threatening “in any manner,” as Justice White would interpret it, to refer only to threats that violate the District’s criminal laws. Post, at 732-733, n. 7. The only threats meeting that definition would have been threats to do physical harm, to kidnap, or to damage property. See D. C. Code Ann. §§22-507, 22-2307 (1989). Threats to stalk, to frighten, to cause intentional embarrassment, to make harassing phone calls, to make false reports to employers or prospective employers, to harass by phone calls or otherwise at work — to mention only a few of the additional threats that might be anticipated in this domestic situation — would not be cov*705ered. Surely “in any manner threaten” should cover at least all threats to commit acts that would be tortious under District of Columbia law (which would be consistent with the trial court’s later reference to a “legal threat”). Thus, under our Blockburger analysis the aggravated threat counts and the assault-with-intent-to-kill count come out the same way.
Justice White attempts to avoid this issue altogether because, in his view, it would be “injudicious” to consider the differences in Foster, not pressed by the Government, between the CPO restrictions and the alleged statutory offenses. Post, at 740. Of course, these differences are pure facts, apparent on the face of the CPO and the indictment. They do not alter the question presented, which assumes only that the prosecuted conduct was the same, see supra, at 694, not that the terms of the CPO and the statute were. Further, although the Government did not argue that the different counts in Foster should come out differently, it did argue (as we do) that they all should be evaluated under Blockburger and not Grady, see, e. g., Brief for United States 14-15, 42; and we are not aware of any principle that prevents us from accepting a litigant’s legal theory unless we agree with the litigant on all the applications of the theory. The standard to be applied in determining the double jeopardy effect of criminal charges based on the same conduct (Blockburger vs. Grady) assuredly is included within the question presented. That makes Justice White’s citation of cases declining to consider legal issues not raised below wholly beside the point. Nor can we see any abuse of what Justice White himself regards as a prudential limitation, when the evident factual difference between the charges and the CPO order is central to proper constitutional analysis.
Justice Souter has apparently been led astray by his misinterpretation of the word “incidents” in the following passage of Nielsen: “[W]here, as in this case, a person has been tried and convicted for a crime which has various incidents included in it, he cannot be a second time tried for one of those incidents without being twice put in jeopardy for the same offence.” 131 U. S., at 188. He apparently takes “incident” to mean “event” or “conduct.” See post, at 752, and n. 5, 757-758. What it obviously means, however, is “element.” See Black’s Law Dictionary 762 (6th ed. 1990) (defining “incidents of ownership”); J. Bouvier, Law Dictionary 783-784 (1883) (defining “incident” and giving examples of “incident to a *708reversion,” and “incidents” to a contract). That is perfectly clear from the very next sentence of Nielsen (which Justice Souter does not quote): “It may be contended that adultery is not an incident of unlawful cohabitation . ...” 131 U. S., at 189.
There is, for example, no other way to read the following passage in Illinois v. Vitale, quoted by Justice Souter, post, at 757: “[In Harris] we treated a killing in the course of a robbery as itself a separate statutory offense, and the robbery as a species of lesser-ineluded offense.” 447 U. S. 410, 420 (1980).
Justice Souter contends that Burton is not in point because the case arose on a demurrer to the indictment, so that the Court “was not presented with the factual basis for the charges.” Post, at 758. It would be a rare and unsatisfactory indictment that did not set forth the factual basis for the charges. The Court in Burton discusses the facts at length. 202 U. S., at 379-381. It is obvious, and it was assumed by the Court, that the same conduct was at issue in both indictments. Having decided, pursuant to Blockburger, that the nature of the statutes did not support a claim of double jeopardy, the Court (if it agreed with Justice Souter’s view of the law) should have proceeded to consider whether the nature of the acts alleged supported such a claim.
Both Justice White, post, at 735, and Justice Souter, post, at 768-759, recognize that Gavieres did hold that Blockburger is the only test for “same offence.” Justice Souter handles this difficulty by simply ignoring the concession. See ibid. Justice White first minimizes the concession, arguing that application of our version of Blockburger to successive prosecutions has happened (by reason of Gavieres) “only once.” Post, at 735. Once, it seems to us, is enough to make a precedent. Justice White then seeks to neutralize the precedent by offering still another case, Grafton v. United States, 206 U. S. 333 (1907), that cannot support the reading grafted onto it today. Post, at 739-740. The defendant in Grafton was first tried and acquitted by a military court for the offense of homicide, and then tried by a civilian criminal court for assassination, and convicted of homicide, based on the same conduct. 206 U. S., at 349. The second prosecution was held barred by the Double Jeopardy Clause. Justice White argues that, just as Grafton had to be a soldier for the military court to have jurisdiction, so too here the only relevance of the CPO is that it gave the court authority to punish offenses “already prescribed by the criminal law.” Post, at 740. This description does not accurately portray the threat counts, see n. 8, supra — but the problem with Justice White’s analysis is deeper than that. The substantive offense for which Grafton was first tried (violation of Philippines Penal Code Article 404) did not have as one of its elements status as a soldier, whereas the substantive offense for which Poster was first tried did have as one of its elements knowledge of an extant CPO. See supra, at 700-702. Since military status was not an element of Grafton’s charged offense, it is not *711true that our analysis would produce a result contrary to the opinion in Grafton. Under the traditional Blockburger elements test, assassination, as defined in Article 403 of the Philippines Penal Code, contained an element that homicide, as defined in Article 404, did not; but, as the Court noted, homicide did not contain any element not included in assassination. 206 U. S., at 350 (“One crime may be a constituent part of the other”); accord, id,., at 355 (he “could not subsequently be tried for the same offense”). Grafton could therefore not later be prosecuted for assassination, much less later be convicted for the very same homicide offense of which he had been acquitted. (In fact, Grafton may simply have been decided on grounds of collateral estoppel, see id., at 349-351, an issue that we specifically decline to reach in this case, see n. 17, infra.)
It is unclear what definition of “same offence” Justice Souter would have us adopt for successive prosecution. At times, he appears content with our having added to Blockburger the Grady same-conduct test. At other times, however, he adopts an ultra-Grady “same transaction” rule, which would require the Government to try together all offenses (regardless of the differences in the statutes) based on one event. See post, at 747, 761. Of course, the same-transaction test, long espoused by Justice Brennan, see, e. g., Brown v. Ohio, 432 U. S. 161, 170 (1977) (concurring opinion), has been consistently rejected by the Court. See, e. g., Garrett v. United States, 471 U. S. 773, 790 (1985).
Justice Souter dislikes this result because it violates “the principles behind the protection from successive prosecutions included in the Fifth Amendment.” Post, at 761. The “principles behind” the Fifth Amendment are more likely to be honored by following longstanding practice than by following intuition. But in any case, Justice Souter’s concern that prosecutors will bring separate prosecutions in order to perfect their case seems unjustified. They have little to gain and much to lose from such a strategy. Under Ashe v. Swenson, 397 U. S. 436 (1970), an acquittal *713in the first prosecution might well bar litigation of certain facts essential to the second one — though a conviction in the first prosecution would not excuse the Government from proving the same facts the second time. Surely, moreover, the Government must be deterred from abusive, repeated prosecutions of a single offender for similar offenses by the sheer press of other demands upon prosecutorial and judicial resources. Finally, even if Justice Souter’s fear were well founded, no double jeopardy bar short of a same-transaction analysis will eliminate this problem; but that interpretation of the Double Jeopardy Clause has been soundly rejected, see, e. g., Garrett, supra, and would require overruling numerous precedents, the latest of which is barely a year old, United States v. Felix, 503 U. S. 378 (1992).
See, e. g., Sharpton v. Turner, 964 F. 2d 1284, 1287 (CA2) (Grady formulation “has proven difficult to apply” and “whatever difficulties we have previously encountered in grappling with the Grady language have not been eased by” Felix), cert. denied, 506 U. S. 986 (1992); Ladner v. Smith, 941 F. 2d 356, 362, 364 (CA5 1991) (a divided court adopts a four-part test for application of Grady and notes that Grady, “even if carefully analyzed and painstakingly administered, is not easy to apply”), cert. denied, 503 U. S. 983 (1992); United States v. Calderone, 917 F. 2d 717 (CA2 1990) (divided court issues three opinions construing Grady), vacated and remanded, 503 U. S. 978 (1992) (remanded for consideration in light of Felix); United States v. Prusan, 780 F. Supp. 1431, 1434-1436 (SDNY 1991) (“[T]he lower courts have had difficulty discerning the precise boundaries of the Grady standard, and the circuits have not applied uniformly the ‘same conduct’ test”), rev’d, 967 F. 2d 57 (CA2), cert. denied sub nom. Vives v. United States, 506 U. S. 987 (1992); State v. Woodfork, 239 Neb. 720, 725, 478 N. W. 2d 248, 252 (1991) (divided court overrules year-old precedent construing Grady, because it was a “misapplication” of Grady); Eatherton v. State, 810 P. 2d 93, 99, 104 (Wyo. 1991) (majority states that *714“[t]he Supreme Court did not really develop any new law In Grady with respect to successive prosecutions,” while dissent concludes that Grady requires reversal). Commentators have confirmed that Grady contributed confusion rather than certainty. See Poulin, Double Jeopardy Protection against Successive Prosecutions in Complex Criminal Cases: A Model, 25 Conn. L. Rev. 95 (1992); Thomas, A Modest Proposal to Save the Double Jeopardy Clause, 69 Wash. U. L. Q. 195 (1991).
We do not address the motion to dismiss the threat counts based on collateral estoppel, see Ashe v. Swenson, supra, because neither lower court ruled on that issue.
Justices White, Stevens, and Souter concur in this portion of the judgment.
Justice Blackmun concurs only in the judgment with respect to this portion.