I concur in the judgment. I agree with the majority’s holding that a noncapital criminal defendant, charged for sentencing purposes with having suffered a prior felony conviction, may not move in the sentencing court to strike the earlier conviction on the ground his or her Boykin-Tahl rights (Boykin v. Alabama (1969) 395 U.S. 238 [89 S.Ct. 1709, 23 L.Ed.2d 274] (Boykin); In re Tahl (1969) 1 Cal.3d 122 [81 Cal.Rptr. 577, 460 P.2d 449] (Tahl)) were violated in the previous case, if the prior conviction predates our decision in Tahl. As the majority explain, when People v. Sumstine (1984) 36 Cal.3d 909 [206 Cal.Rptr. 707, 687 P.2d 904] (Sumstine) authorized such Boykin-Tahl motions, it relied on the premise that disruption of the current proceeding would be minimal, because a Boykin-Tahl violation would usually appear on the face of the plea record in the prior case. However, only after Tahl were courts on notice that “the record [of a guilty plea] must contain on its face” direct evidence of the required pre-plea admonitions and waivers. (Tahl, supra, 1 Cal.3d at p. 132, italics in original.) Thus, even if motions to strike post -Tahl priors can often be handled with reasonable dispatch, such is not the case for pleas taken prior to Tahl. In the latter category of cases, determination of the knowing and voluntary nature of the prior plea “would be an onerous task” (maj. opn., ante, at p. 443) which often could not be accomplished by mere quick reference to the face of the prior record. Permitting Sumstine challenges to pre-Tahl guilty pleas would therefore “be judicially inefficient and [would] saddle the trial courts with an unreasonable burden.” (Ibid.)
Because the prior conviction at issue in this case occurred on January 28, 1969, more than nine months before we decided Tahl, it cannot be challenged by a Sumstine motion. The trial court’s denial of the motion was therefore correct. We need decide nothing more to dispose of this case.
Nonetheless, the majority go out of their way to affirm that Sumstine, a decision compelled by neither the federal nor the state Constitution, otherwise still entitles a defendant to challenge a charged prior conviction on Boykin-Tahl grounds by means of a motion to strike. The majority reach this result despite intervening decisions that cast serious doubt on Sumstine’s reasoning.
*445As the majority acknowledge, after Sumstine was decided, Custis v. United States (1994) 511 U.S. 485 [114 S.Ct. 1732, 128 L.Ed.2d 517] (Custis) held that in federal recidivism proceedings, a prior conviction can be challenged only on grounds the defendant neither received nor waived representation by counsel. (See Gideon v. Wainwright (1963) 372 U.S. 335 [83 S.Ct. 792, 9 L.Ed.2d 799]; see also Burgett v. Texas (1967) 389 U.S. 109 [88 S.Ct. 258, 19 L.Ed.2d 319].) Among other things, Custis specifically declined to allow such challenges on grounds that a prior guilty plea was taken without a proper advisement and waiver of rights. In direct contradiction to the analysis of Sumstine, Custis concluded that both administrative considerations, and the strong policy favoring finality of criminal judgments, counseled against such a course. (Custis, supra, 511 U.S. at pp. 496-497 [114 S.Ct. at p. 1739].)
Less than three years ago, we found Custis persuasive enough to conclude that in noncapital cases at least, a Sumstine motion may not be used to challenge a prior conviction on grounds of ineffective assistance of counsel. (Garcia v. Superior Court (1997) 14 Cal.4th 953 [59 Cal.Rptr.2d 858, 928 P.2d 572] (Garcia).) In particular, we stressed the disruption of orderly proceedings that would occur if recidivist defendants were allowed to encumber their trials for later offenses by demanding lengthy examination of the circumstances under which prior charged convictions were obtained. (Id., at pp. 964-966.)
Were we writing on a clean slate, I would follow Custis here as well, and would therefore decline to recognize, in noncapital cases, a motion to strike a prior conviction on any ground other than the complete denial of counsel. I discuss below some of the administrative difficulties of allowing such a motion, even as limited to Boykin-Tahl issues. But even aside from the administrative problems, I sympathize with Custis’’ s independent concern for the finality of judgments.
As Custis explained, “ ‘[i]nroads on the concept of finality tend to undermine confidence in the integrity of our procedures’ and inevitably delay and impair the orderly administration of justice. [Citation.] [The] principles of finality associated with habeas corpus actions apply with at least equal force when a defendant seeks to attack a previous conviction used for sentencing. By challenging the previous conviction, the defendant is asking a [trial] court ‘to deprive [the] [prior] judgment of [its] normal force and effect in a proceeding that ha[s] an independent purpose other than to overturn the prior judgmenft].’ [Citation.] These principles bear extra weight in cases in which the prior convictions . . . are based on guilty pleas, because when a guilty plea is at issue, ‘the concern with finality served by the limitation on collateral *446attack has special force.’ [Citation.]” (Custis, supra, 511 U.S. 485, 497 [114 S.Ct. 1732, 1739], italics added.)
One might nonetheless conclude that Custis does not justify entirely overruling our preexisting precedent in Sumstine. However, since the majority take this opportunity to reaffirm the availability of a Sumstine motion for Boykin-Tahl claims, Custis should at least inspire the majority to limit the motion carefully, so that administrative disruptions and intrusions on the finality of judgments are minimized. The majority have wisely adopted one such limitation by making clear that a Sumstine motion cannot be employed to attack California prior convictions that predate Tahl. Other clarifications of the reach of Sumstine, not explicitly endorsed by the majority, come readily to mind.
First, just as a California conviction that predates Tahl cannot be the subject of a Sumstine motion on Boykin-Tahl grounds, out-of-state convictions should similarly be excluded, at least absent a clear showing they arose under guilty plea requirements identical to those imposed in this state by Tahl. To be sure, it has long been plain error under Boykin for any American court to take a plea of guilt without an “affirmative showing” on the record that the plea was voluntary, with a full understanding of the fundamental constitutional trial rights thereby forfeited. (Boykin, supra, 395 U.S. 238, 242 [89 S.Ct. 1709, 1711-1721].) But Boykin did not explicitly require that full advisement of the defendant, and the defendant’s subsequent waivers, must themselves occur on the record. In California, Tahl imposed these stricter prophylactic requirements in order to assure adherence to Boykin and to foreclose any possible dispute about the validity of a particular plea. (Tahl, supra, 1 Cal.3d 122, 132.) It is Tahl’s procedural exactitude, in particular, on which Sumstine and the instant majority have heavily relied to conclude that motions to strike prior convictions on Boykin-Tahl grounds do not impose excessive burdens on sentencing courts.
I do not know the extent to which other jurisdictions have interpreted Boykin to require the, facial admonitions and waivers prescribed for California courts by Tahl. However, in People v. Howard (1992) 1 Cal.4th 1132 [5 Cal.Rptr.2d 268, 824 P.2d 1315], we noted the overwhelming weight of federal authority is to the effect that Boykin does not require explicit on-the-record admonitions and waivers, and allows a guilty plea to be upheld, even on direct appeal, if the record as a whole affirmatively shows the plea was knowing and voluntary. (Howard, supra, at pp. 1177-1179, citing, inter alia, Brady v. United States (1970) 397 U.S. 742, 747-748, fn. 4 [90 S.Ct. 1463, 1468, 25 L.Ed.2d 747]; United States v. Pricepaul (9th Cir. 1976) 540 F.2d 417, 424-425.)
*447The majority acknowledge that the Sumstine rule, as applied to Boykin-Tahl issues, is tolerable only insofar as we can expect the record of the challenged prior guilty plea readily to show, on its face, that the defendant knew and waived his rights. For this reason, only priors governed by Tahl’s requirement of express admonitions and waivers may be the subject of a Sumstine motion. Just as this principle eliminates Boykin-Tahl challenges to California priors that predate Tahl, so must Boykin-Tahl challenges to non-California priors be excluded, except where it appears beyond doubt that the guilty pleas underlying such convictions were subject, under the law of the convicting jurisdictions, to TaM-like procedural formalities.
Second, we should emphasize a point that seems obvious, but has not been made crystal clear by Sumstine or by the instant majority opinion. In a noncapital case at least, if the available record of the prior conviction directly reflects the explicit admonitions and waivers required by Tahl, or if the evidence otherwise shows such explicit admonitions and waivers occurred in open court, a Sumstine challenge should be summarily denied, without any need for further proceedings.
This principle should bar “behind the record” challenges to a facially valid waiver proceeding, such as claims that the defendant was not mentally competent at the time he received and waived his rights. When a defendant received the explicit admonitions contemplated by Tahl, and then stated, under the direct observation of court and counsel, that he understood the consequences of his plea and voluntarily waived his rights, his belated protestations to the contrary should not be allowed to affect the orderly disposition of a subsequent criminal proceeding. To allow such disputes in the face of the prior record would create exactly the kind of unwarranted midtrial disruptions that influenced the holdings of both Custis and Garcia.1
Finally, the administrative concerns expressed in Custis and Garcia should cause us to examine more closely the implications of an incomplete record, even in the case of a plea governed by the explicit procedural requirements of Tahl.2 Both Sumstine and the instant majority opinion rely heavily on the assumption that a complete record of the prior conviction will be readily *448available, so that it can easily be ascertained whether Tahl’s procedures were followed. However, our own decisions belie this assumption.
Thus, in Sumstine itself, we noted that even though the prior guilty plea there at issue occurred in 1974, long after Boykin and Tahl, the documentary record was “woefully inadequate” to determine whether the requirements of that case had been met. (Sumstine, supra, 36 Cal.3d 909, 921, 924.) We observed that a sufficient record might have existed at one time, but reporter’s notes of the 1974 arraignment had apparently been eliminated under a statute that permitted county clerks to destroy such notes after a specified period. (Id., at p. 921.)
Sumstine gave the following explanation of how a motion to strike should proceed in such circumstances: The defendant cannot rely on the silence of the prior record, but must “allege” the actual denial of his constitutional rights. Moreover, once the prosecution proves the existence of the prior conviction, the burden shifts to the defendant to “produc[e] evidence” that his rights were infringed. If the defendant does so, the prosecution may introduce rebuttal evidence, but at this stage, “it will not be sufficient rebuttal for the state to simply invoke the regularity of the silent record.” (Sumstine, supra, 36 Cal.3d 909, 923; see also People v. Coffey (1967) 67 Cal.2d 204, 217 [60 Cal.Rptr. 457, 430 P.2d 15].)
In Curl v. Superior Court (1990) 51 Cal.3d 1292 [276 Cal.Rptr. 49, 801 P.2d 292] (Curl), we held that a capital defendant could move to strike, on grounds his guilty plea was involuntary, a prior murder conviction alleged as a special circumstance. However, stressing that a judgment under collateral attack is presumed regular (Evid. Code, §§ 664, 666), and that the party with the burden of proof of a fact has the initial duty to produce evidence of that fact (id., § 550, subd. (b)), Curl concluded that the defendant attacking a prior conviction has the burden to prove its constitutional invalidity by a preponderance of evidence. (Curl, supra, 51 Cal.3d at pp. 1303-1306.)
In Curl itself, “[n]o transcript of the . . . [guilty plea] proceedings [in the 1977 prior murder case] was available.” (Curl, supra, 51 Cal.3d 1292, 1296.) The 1977 minute order merely stated that court and counsel had “voir dire[d]” defendant on the charge, and that the court “ ‘[found] the pleas to be free and voluntary . . . with an intelligent waiver of rights and with understanding of the nature of the charges and possible consequences . . . .’” (Ibid.)
Despite these available (if incomplete) indications that the defendant had received and waived his rights, a Sumstine evidentiary hearing was held. The *449defendant testified he was on drags at the time of the 1977 plea, misunderstood its penal consequences, was not told the rights he was waiving, and would not have pled guilty if fully advised. (Curl, supra, 51 Cal.3d 1292, 1297-1298.) The judge and defense counsel in the 1977 case gave rebuttal testimony suggesting it was their “custom and habit” to assure fully informed waivers by a defendant who seemed competent. (Ibid.) The 1977 prosecutor testified he specifically remembered Boykin-Tahl admonitions and waivers, had seen no outward signs of intoxication, and kept file notes suggesting the defendant was questioned about recent drag ingestion. (Curl, supra, at p. 1297.) On this state of the record, Curl implied, the trial court, which had found “clear and convincing” evidence to rebut any inference of the prior’s invalidity (id., at p. 1298), would necessarily and properly have found the defendant failed to prove his claim by a preponderance of evidence.
In People v. Pride (1992) 3 Cal.4th 195 [10 Cal.Rptr.2d 636, 833 P.2d 643] (Pride), the defendant sought, on Boykin-Tahl grounds, to prevent use of his 1972 rape conviction as aggravating evidence at the penalty phase of his trial for capital murder. No reporter’s transcript of the 1972 guilty plea existed. The minute order recording the plea said the defendant had “waived trial/jury,” but the order did not refer to any other Boykin-Tahl advisements or waivers. An evidentiary hearing was held. There, it developed that Judge Calcagno, who took the 1972 plea, had since died. Both the prosecutor and the principal public defender in the 1972 case took the stand. Neither remembered the individual circumstances of the plea, though both testified it was Judge Calcagno’s invariable practice to provide full advisements, and to extract full waivers, before accepting guilty pleas. (Pride, supra, 3 Cal.4th at pp. 255-256.) The defendant testified he “did not remember” anybody advising him of his constitutional rights before he pled guilty. (Id., at p. 255.) On this record, we concluded, as in Curl, that the evidence of judicial “habit and custom” supported the trial court’s determination that the challenged plea was taken in accordance with Boykin and Tahl. (Pride, supra, at p. 256.)
These complicated scenarios suggest that for multiple, often legitimate, reasons, complete records of long-past guilty pleas, even those taken after Boykin and Tahl, may not be available. The question arises whether, in such circumstances, a noncapital defendant can obtain a full evidentiary hearing, with its attendant disruption of the current trial, merely by “alleging]” (see Sumstine, supra, 36 Cal.3d 909, 923) that his prior plea violated Boykin-Tahl procedures, and that he actually did not know and waive his rights.
In addressing that issue, three principles seem paramount. The first, as was emphasized in both Sumstine and Curl, involves the related presumptions, for purposes of collateral attack, that final judgments are valid, that *450official duty was performed, and that proceedings were regular. Application of these rules means we must assume, until the contrary appears, that court proceedings were conducted in compliance with applicable rules, including those set forth in Tahl. Thus, as Sumstine implied, the defendant mounting a collateral attack on his prior conviction cannot merely invoke the “silence” of the record in the prior case. (36 Cal.3d at p. 924.) Instead, as Curl made clear, he has the burden of proving an affirmative violation of his constitutional rights.
The second principle is that the People should suffer no undue penalty for the inadequacy or unavailability of an earlier record. It is the defendant’s obligation to ensure a record adequate to evaluate his collateral attack on a presumptively valid final judgment. If the defendant never sought to marshal the record in the prior matter, and never timely challenged the prior judgment, the government may well have been within its rights to cull and destroy its obsolete files. The state has no duty to generate “appeal-worthy” records in every case, or to maintain criminal case files indefinitely, on the chance they will one day be relevant on collateral attack in another matter.
The final principle, derived from the reasoning of both Custis and Garcia, is that a noncapital defendant’s opportunity, in a later case, to prove the invalidity of a prior conviction must be balanced against the inherent disruption such a procedure imposes upon the orderly administration of justice. This principle counsels against allowing a noncapital defendant to obtain a full evidentiary hearing, such as those conducted in Pride and Curl, on the basis of a wholly uncorroborated allegation that a prior plea, taken in California after Boykin and Tahl, was actually uninformed or involuntary.
Accordingly, I submit, if a noncapital Sumstine defendant cannot supply a record of the prior plea sufficient to determine whether Boykin-Tahl procedures were followed, he or she must do more than allege an infringement of Boykin-Tahl rights in order to obtain a full evidentiary hearing on that issue. Instead, to avoid immediate denial of such a motion, the defendant should be required to make a prima facie showing (by appropriate declarations, documents, or formal offers of proof) that Boykin-Tahl procedures were violated, and that there was, in fact, no knowing and voluntary waivers of rights.
In making this preliminary showing, the defendant should not be permitted to rely solely on the absent or incomplete prior record and his or her own insistence the prior plea was constitutionally defective. In this context, “a defendant’s self-serving statement [should be] insufficient. . . and [should] be corroborated by independent evidence. A contrary [rule] would lead to an unchecked flow of easily fabricated claims.” (In re Alvernaz (1992) 2 Cal.4th *451924, 938 [8 Cal.Rptr.2d 713, 830 P.2d 747] [imposing similar requirements on a defendant who seeks to show his rejection of a plea offer was the result of ineffective assistance].)3
These limitations all seem necessary to keep the Sumstine procedure within rational bounds, and to prevent defendants, now additionally motivated by three strikes consequences, from delaying current criminal trials with specious but time-consuming attacks on prior guilty pleas. Assuming a narrow version of the Sumstine rule should be retained, I would explicitly circumscribe Sumstine’s availability and procedures accordingly.
Chin, J., and Brown, J., concurred.
Appellant’s petition for a rehearing was denied October 6, 1999.
Thus, proof that the defendant in fact did not understand and voluntarily waive his rights in the prior case, though necessary to his or her final success on a Sumstine motion, should be irrelevant unless and until the defendant first establishes that the proceedings leading to his plea did not comply with Tahl.
By my use of the word “incomplete” in this context, I do not mean a full record that demonstrates the inadequacy or incompleteness of the Boykin-Tahl proceedings conducted in the prior case. Instead, as will become clear, I refer to a record, some or all or which was never prepared or has since been lost or destroyed, such that it does not disclose whether adequate admonitions and waivers occurred in open court.
My comments throughout this opinion relate only to noncapital Sumstine motions that challenge prior convictions on Boykin-Tahl grounds. Because a Sumstine motion that alleges the complete denial of counsel in a prior case is constitutionally based (see Custis, supra, 511 U.S. 485, 495 [114 S.Ct. 1732, 1738]), we may have less latitude to dictate the procedures governing such a motion. I therefore express no views on that subject. Furthermore, by my various references to “noncapital” defendants, I do not mean to imply that I necessarily accept any broader rules or grounds for a Sumstine motion in capital cases, other than as specifically provided in People v. Horton (1995) 11 Cal.4th 1068, 1126-1141 [47 Cal.Rptr.2d 516, 906 P.2d 478],