In The

Supreme Court of the United States




Decided March 7, 2005

Justice O’Connor, Dissenting

Topic: Criminal Procedure*Court vote: 5–3
Click any Justice for detail
Joining O'Connor opinion: Justice BREYER Justice BREYER Justice KENNEDY Justice KENNEDY
Citation: 544 U.S. 13 Docket: 03–9168Audio: Listen to this case's oral arguments at Oyez

* As categorized by the Washington University Law Supreme Court Database

Next opinion >< Previous opinion

DISCLAIMER: Only United States Reports are legally valid sources for Supreme Court opinions. The text below is provided for ease of access only. If you need to cite the exact text of this opinion or if you would like to view the opinions of the other Justices in this case, please view the original United States Report at the Library of Congress or Justia. The Sandra Day O'Connor Institute does not in any way represent, warrant, or guarantee that the text below is accurate."


Justice O’Connor, with whom Justice Kennedy and Justice Breyer join, dissenting.

The Court today adopts a rule that is not compelled by statute or by this Court’s precedent, that makes little sense as a practical matter, and that will substantially frustrate Congress’ scheme for punishing repeat violent offenders who violate federal gun laws. The Court is willing to acknowledge that the petitioner’s prior state burglary convictions occurred, and that they involved unpermitted entries with intent to commit felonies. But the Court refuses to accept one additional, commonsense inference, based on substantial documentation and without any evidence to the contrary: that petitioner was punished for his entries into buildings.

The petitioner, Reginald Shepard, has never actually denied that the prior crimes at issue were burglaries of buildings. Nor has he denied that, in pleading guilty to those crimes, he understood himself to be accepting punishment for burglarizing buildings. Instead, seeking to benefit from the unavailability of certain old court records and from a minor ambiguity in the prior crimes’ charging documents, petitioner asks us to foreclose any resort to the clear and uncontradicted background documents that gave rise to and supported his earlier convictions.

The Court acquiesces in that wish and instructs the federal courts to ignore all but the narrowest evidence regarding an Armed Career Criminal Act defendant’s prior guilty pleas. I respectfully dissent.


The Armed Career Criminal Act of 1984 (ACCA) mandates a 15-year minimum sentence for certain federal firearms violations where the defendant has three prior convictions for a “violent felony.” 18 U. S. C. §924(e). In defining violent felonies for this purpose, Congress has specified that the term includes any crime, punishable by more than one year’s imprisonment, that “is burglary, arson, or extortion, involves use of explosives, or otherwise involves conduct that presents a serious potential risk of physical injury to another.” §924(e)(2)(B)(ii). We held in Taylor v. United States, 495 U. S. 575 (1990), that the statute’s use of the term “burglary” was meant to encompass only what we described as “generic” burglary, a crime with three elements: (i) “unlawful or unprivileged entry into, or remaining in,” (ii) “a building or structure,” (iii) “with intent to commit a crime.” Id., at 598–599.

That left the problem of how to determine whether a defendant’s past conviction qualified as a conviction for generic burglary. The most formalistic approach would have been to find the ACCA requirement satisfied only when the statute under which the defendant was convicted was one limited to “generic” burglary. But Taylor wisely declined to follow that course. The statutes which some States—like Massachusetts here, or Missouri in Taylor —use to prosecute generic burglary are overbroad for ACCA purposes: They are not limited to “generic” burglary, but also punish the nongeneric kind. Restricting the sentencing court’s inquiry to the face of the statute would have frustrated the purposes of the ACCA by allowing some violent recidivists convicted of federal gun crimes to escape the ACCA’s heightened punishment based solely on the fortuity of where they had committed their previous crimes.

Instead, Taylor adopted a more “pragmatic” approach. Ante, at 6 (majority opinion). Every statute punishes a certain set of criminalized actions; the problem with some burglary statutes, for purposes of the ACCA, is that they are overinclusive. But Taylor permitted a federal court to “go beyond the mere fact of conviction”—and to determine, by using other sources, whether the defendant’s prior crime was in the subset of the statutory crime qualifying as generic burglary. For example, where a defendant’s prior conviction occurred by jury trial, Taylor instructed the federal court to review “the indictment or information and jury instructions” from the earlier conviction, to see whether they had “required the jury to find all the elements of generic burglary in order to convict.” 495 U. S., at 602.

As the Court recognizes, however, Taylor ’s use of that one example did not purport to be exhaustive. See ante, at 6. See also United States v. Harris, 964 F. 2d 1234, 1236 (CA1 1992) (Breyer, C. J.). Rather, Taylor left room for courts to determine which other reliable and simple sources might aid in determining whether a defendant had in fact been convicted of generic burglary. The Court identifies several such sources that a sentencing judge may consult under the ACCA: the “charging document, written plea agreement, transcript of plea colloquy, and any explicit factual finding by the trial judge to which the defendant assented.” Ante, at 1–2. I would expand that list to include any uncontradicted, internally consistent parts of the record from the earlier conviction. That would include the two sources the First Circuit relied upon in this case.

Shepard’s four prior convictions all occurred by guilty pleas to charges under Massachusetts’ two burglary statutes—statutes that punish “[w]hoever … breaks and enters a building, ship, vessel or vehicle, with intent to commit a felony.” Mass. Gen. Laws Ann., ch. 266, §16 (West 2000) (emphasis added); see also §18. The criminal complaints used as charging documents for the convictions at issue did not specify that Shepard’s offenses had involved a building, but instead closely copied the more inclusive language of the appropriate statute. If these complaints were the only evidence of the factual basis of Shepard’s guilty pleas, then I would agree with the majority that there was no way to know whether those convictions were for burglarizing a building. But the Government did have additional evidence. For each of the convictions, the Government had both the applications by which the police had secured the criminal complaints and the police reports attached to those applications. Those documents decisively show that Shepard’s illegal act in each prior conviction was the act of entering a building. Moreover, they make inescapable the conclusion that, at each guilty plea, Shepard understood himself to be admitting the crime of breaking into a building.

Consider, for instance, the first burglary conviction at issue. The complaint for that conviction alleged that, on May 6, 1989, Shepard “did break and enter in the night time the building, ship, vessel or vehicle, the property of Jerri Cothran, with intent to commit a felony therein” in violation of §16. 3 App. 5. The place of the offense was alleged as “30 Harlem St.,” and the complaint contained a cross-reference to “CC#91–394783.”

The majority would have us stop there. Since both the statute and the charging document name burglary of a “building, ship, vessel or vehicle,” the majority concludes that there is no way to tell whether Massachusetts punished Shepard for transgressing its laws by burglarizing a building, or for doing so by burglarizing a vehicle, ship, or vessel. (Although the majority would also allow a look at Shepard’s written plea agreement or a transcript of the plea proceedings, those items are no longer available in Shepard’s case, since Massachusetts has apparently seen little need to preserve the miscellany of long-past convictions.)

I would look as well to additional portions of the record from that plea—the complaint application and police report. The complaint application lists the same statute, describes (in abbreviated form) the same offense, names the same victim and address, and contains the same reference number (though differently hyphenated) as the complaint itself. In addition, the application specifies as relevant “PROPERTY” (meaning “Goods stolen, what destroyed, etc.”) a “Cellar Door.” Id., at 6. The police report (which also names the same victim, date, and place of offense, and contains the same reference number as the other two documents) gives substantially more detail about why Massachusetts began criminal proceedings against Shepard:

“[R]esponded to [radio call] to 30 Harlem St. for B&E; in progress. On arrival observed cellar door in rear had been broken down. Spoke to victim who stated that approx 3:00 a.m. she heard noises downstairs. She then observed suspect … in her pantry.” Id., at 7.

Three points need to be made about the relationship between the complaint (whose use the majority finds completely unobjectionable) and the application and police report (which I would also consider). First, all of the documents concern the same crime. Second, the three documents are entirely consistent—nothing in any of them casts doubt on the veracity of the others. Finally, and most importantly, the common understanding behind all three documents was that, whatever the range of conduct punishable by the state statute, this defendant was being prosecuted for burglary of a building. See 348 F. 3d 308, 314 (CA1 2003) (“[T]here is a compelling inference that the plea was to the complaint and that the complaint embodied the events described in the application or police report in the case file”).

There certainly is no evidence in the record contradicting that understanding. Notably, throughout these proceedings, Shepard has never denied that the four guilty pleas at issue involved breaking into buildings. Nor has he denied that his contemporaneous understanding of each plea was that, as a result of his admission, he would be punished for having broken into a building. During his federal sentencing hearings, Shepard did submit an affidavit about his prior convictions. But that affidavit carefully dances around the key issues of what Shepard actually did to run afoul of the law and what he thought was the substance of his guilty plea. Rather, the affidavit focuses on what the judge said to Shepard at the hearing and what Shepard said in response. Even in that regard, the affidavit is strangely ambiguous. In discussing the first conviction, for instance, the affidavit states that “the judge [who took the plea] did not read” the police report to Shepard, “and did not ask me whether or not the information contained in the … report was true.” 1 App. 100. See also ibid. (“I did not admit the truth of the information contained in the … report as part of my plea and I have never admitted in court that the facts alleged in the reports are true”). The affidavit’s statements about the other three prior convictions are similar.

Those statements could be taken as Shepard’s denial that he was ever asked about (or ever admitted to) any of the specific facts of his crime that happen to be mentioned in the police reports—facts like the date and place of the offense, whether he entered through a cellar door and proceeded to the pantry, and so on. But to believe that, we would have to presume that all four Massachusetts courts violated their duty under state law to ensure themselves of the factual basis for Shepard’s plea. In Massachusetts, “[a] defendant’s choice to plead guilty will not alone support conviction; the defendant’s guilt in fact must be established.” Commonwealth v. DelVerde, 398 Mass. 288, 296, 496 N. E. 2d 1357, 1362 (1986). As a result, even if “the defendant admits to the crime in open court, … a court may not convict unless there are sufficient facts on the record to establish each element of the offense.” Id., at 297, 496 N. E. 2d, at 1363. See also Commonwealth v. Colon, 439 Mass. 519, 529, n. 13, 789 N. E. 2d 566, 573, n. 13 (2003) (guilty plea requires admission to the facts); 2 E. Blumenson, S. Fisher, & D. Kanstroom, Massachusetts Criminal Practice §37.7B, p. 288 (1998) (“Usually this is accomplished by the recitation of either the grand jury minutes or police reports, but defendant’s admissions during the plea, or trial evidence, can also support the factual basis” (footnote omitted)). Cf. Commonwealth v. Forde, 392 Mass. 453, 458, 466 N. E. 2d 510, 513 (1984) (conviction cannot be based on uncorroborated confession; rather, there must be some evidence that the crime was “real and not imaginary”). It is thus unlikely that Shepard really intended his affidavit as a statement that none of the various facts found in the police reports were ever admitted by him or discussed in his presence during his guilty pleas.

More likely, Shepard’s attorney carefully phrased the affidavit so that it would admit of a different meaning: that the plea courts never asked, and Shepard never answered, the precise question: “Is what the police report says true?” But I fail to see how that is relevant, so long as Shepard understood that, in pleading guilty, he was agreeing to be punished for the building break-in that was the subject of the entire proceeding.

There may be some scenarios in which—as the result of charge bargaining, for instance, or due to unexpected twists in an investigation—a defendant’s guilty plea is premised on substantially different facts than those that were the basis for the original police investigation. In such a case, a defendant might well be confused about the practical meaning of his admission of guilt. Cf. Taylor, 495 U. S., at 601–602 (“[I]f a guilty plea to a lesser, nonburglary offense was the result of a plea bargain, it would seem unfair to impose a sentence enhancement as if the defendant had pleaded guilty to burglary”). But there is no claim of such circumstances here: All signs are that everyone involved in each prior plea—from the judge, to the prosecutor, to the defense lawyer, to Shepard himself—understood each plea as Shepard’s admission that he had broken into the building where the police caught him. Given each police report’s never-superseded allegation that Shepard had burglarized a building, it strains credulity beyond the breaking point to assert that, in each case, Shepard was actually prosecuted for and pleaded guilty to burglarizing a ship or a car. The lower court was surely right to detect “an air of make-believe” about Shepard’s case. 348 F. 3d, at 311.

The majority’s rule, which forces the federal sentencing court to feign agnosticism about clearly knowable facts, cannot be squared with the ACCA’s twin goals of incapacitating repeat violent offenders, and of doing so consistently notwithstanding the peculiarities of state law. Cf. Taylor, supra, at 582 (“ ‘[I]n terms of fundamental fairness, the Act should ensure, to the extent that it is consistent with the prerogatives of the States in defining their own offenses, that that same type of conduct is punishable on the Federal level in all cases’ ” (quoting S. Rep. No. 98–190, p. 20 (1983))). The Court’s overscrupulous regard for formality leads it not only to an absurd result, but also to a result that Congress plainly hoped to avoid.


The Court gives two principal reasons for today’s ruling: adherence to the Court’s decision in Taylor, and constitutional concerns about the defendant’s right to a jury trial.

The first is hardly convincing. As noted above, Taylor itself set no rule for guilty pleas, and its list of sources for a sentencing court to consider was not intended to be exhaustive. Supra, at 3. The First Circuit’s disposition of this case, therefore, was not in direct conflict with Taylor. Nor did it conflict with the spirit of Taylor. Taylor was in part about “[f]air[ness]” to defendants. 495 U. S., at 602. But there is nothing unfair (and a great deal that is positively just) about recognizing and acting upon plain and uncontradicted evidence that a defendant, in entering his prior plea, knew he was being prosecuted for and was pleading guilty to burglary of a building. Taylor also sought to avoid the impracticality of mini-sentencing-trials featuring opposing witnesses perusing lengthy transcripts of prior proceedings. Id., at 601. But no such problem presents itself in this case: The Government proposed using only the small documentary record behind Shepard’s pleas. Those documents relate to facts that Shepard does not dispute, and Shepard has not indicated any desire to submit counterevidence.

The issue most central to Taylor was the need to effectuate Congress’ “categorical approach” to sentencing recidivist federal offenders—an approach which responds to the reality of a defendant’s prior crimes, rather than the happenstance of how those crimes “were labeled by state law.” Id., at 589. But rather than promote this goal, the majority opinion today injects a new element of arbitrariness into the ACCA: A defendant’s sentence will now depend not only on the peculiarities of the statutes particular States use to prosecute generic burglary, but also on whether those States’ record retention policies happen to preserve the musty “written plea agreement[s]” and recordings of “plea colloqu[ies]” ancillary to long-past convictions. Ante, at 1. In other words, with respect to this most critical issue, the majority’s rule is not consistent with Taylor at all.

That is why I strongly suspect that the driving force behind today’s decision is not Taylor itself, but rather “[d]evelopments in the law since Taylor.” Ante, at 9. A majority of the Court defends its rule as necessary to avoid a result that might otherwise be unconstitutional under Apprendi v. New Jersey, 530 U. S. 466 (2000), and related cases. Ante, at 10–12 (plurality opinion); ante, at 2–3 (Thomas, J., concurring in part and concurring in judgment). I have criticized that line of cases from the beginning, and I need not repeat my reasoning here. See Id., at 523 (dissenting opinion); Ring v. Arizona, 536 U. S. 584, 619 (2002) (dissenting opinion); Blakely v. Washington, 542 U. S. ___, ___ (2004) (slip op., at 8–10) (dissenting opinion). See also Jones v. United States, 526 U. S. 227, 254 (1999) (Kennedy, J., dissenting); Blakely, supra, at ___ (slip op., at 13–17) (Breyer, J., dissenting); United States v. Booker, 543 U. S. ___, ___ (2005) (slip op., at 2–6) (Breyer, J., dissenting). It is a battle I have lost.

But it is one thing for the majority to apply its Apprendi rule within that rule’s own bounds, and quite another to extend the rule into new territory that Apprendi and succeeding cases had expressly and consistently disclaimed. Yet today’s decision reads Apprendi to cast a shadow possibly implicating recidivism determinations, which until now had been safe from such formalism. See Blakely, supra, at ___ (slip op., at 5) (“ ‘ Other than the fact of a prior conviction, any fact that increases the penalty of a crime beyond the prescribed statutory maximum must be submitted to a jury, and proved beyond a reasonable doubt’ ”) (emphasis added; quoting Apprendi, supra, at 490)). See also Booker, supra, at ___ (slip op., at 20) (opinion of the Court by Stevens, J.) (similar).

Even in a post Apprendi world, I cannot understand how today’s case raises any reasonable constitutional concern. To the contrary, this case presents especially good reasons for respecting Congress’ long “tradition of treating recidivism as a sentencing factor” determined by the judge, Almendarez-Torres v. United States, 523 U. S. 224, 243 (1998), rather than as a substantive offense element determined by the jury. First, Shepard’s prior convictions were themselves “established through procedures satisfying the fair notice, reasonable doubt, and jury trial guarantees.” Jones, supra, at 249. Second, as with most recidivism determinations, see Almendarez-Torres, supra, at 235, the burglary determination in Shepard’s case concerned an extremely narrow issue, with the relevant facts not seriously contested. See supra, at 6–7 (discussing shortcomings of Shepard’s affidavit). Finally, today’s hint at extending the Apprendi rule to the issue of ACCA prior crimes surely will do no favors for future defendants in Shepard’s shoes. When ACCA defendants in the future go to trial rather than plead guilty, the majority’s ruling in effect invites the Government, in prosecuting the federal gun charge, also “to prove to the jury” the defendant’s prior burglaries. Almendarez-Torres, 523 U. S., at 234–235. “[T]he introduction of evidence of a defendant’s prior crimes risks significant prejudice,” id., at 235, and that prejudice is likely to be especially strong in ACCA cases, where the relevant prior crimes are, by definition, “violent,” 18 U. S. C. §924(e). In short, whatever the merits of the Apprendi doctrine, that doctrine does not currently bear on, and should not be extended to bear on, determinations of a defendant’s past crimes, like the ACCA predicates at issue in Shepard’s case. The plurality’s concern about constitutional doubt, ante, at 10–12, and Justice Thomas’ concern about constitutional error, ante, at 2–3, are therefore misplaced.


For the reasons explained above, I would find that the First Circuit properly established the applicability of the ACCA sentence by looking to the complaint applications and police reports from the prior convictions. Because the Court concludes otherwise, I respectfully dissent.

Supreme Court icon marking end of opinion

Header photo: United States Supreme Court. Credit: Patrick McKay / Flickr - CC.