I concur in the affirmance of the judgment as to both guilt and penalty. I write separately to explain my reasons for concluding that the guilt verdicts should be affirmed even though the trial court erred in instructing the jury that it could consider evidence of the Rambo murder for “any purpose,” including defendant’s “character or any trait of his character.”
In assessing the prejudice to a criminal defendant from an error that permits the jury to consider the defendant’s uncharged acts as evidence of *211character or criminal propensity, this court and the Courts of Appeal have consistently applied the harmless error standard required by our state Constitution (Cal. Const., art. VI, § 13) and articulated in People v. Watson (1956) 46 Cal.2d 818, 836 [299 P.2d 243]. (See, e.g., People v. Miranda (1987) 44 Cal.3d 57, 83 [241 Cal.Rptr. 594, 744 P.2d 1127]; People v. Alcala (1984) 36 Cal.3d 604, 636 [205 Cal.Rptr. 775, 685 P.2d 1126]; People v. Dominguez (1981) 121 Cal.App.3d 481, 501 [175 Cal.Rptr. 445]; People v. Moran (1973) 33 Cal.App.3d 724, 729 [109 Cal.Rptr. 287].) Defendant urges this court to depart from this consistent practice of applying the state harmless error standard and to apply instead the federal standard as described in Chapman v. California (1967) 386 U.S. 18, 24 [17 L.Ed.2d 705, 710-711, 87 S.Ct. 824, 24 A.L.R.3d 1065], and further explained in Yates v. Evatt (1991) 500 U.S. 391 [114 L.Ed.2d 432, 111 S.Ct. 1884]. Defendant maintains that the federal standard is required because the error denied him the due process of law guaranteed by the Fifth and Fourteenth Amendments to the United States Constitution.
Defendant cites no holding under the federal Constitution prohibiting jury consideration of uncharged crimes to prove criminal propensity. Indeed, the United States Supreme Court just two years ago declined to express an opinion “on whether a state law would violate the Due Process Clause if it permitted the use of ‘prior crimes’ evidence to show propensity to commit a charged crime.” (Estelle v. McGuire (1991) 502 U.S. _, _, fn. 5 [116 L.Ed.2d 385, 401, 112 S.Ct. 475, 484, fn. 5].) Nevertheless, defendant argues that the instruction in this case denied him due process of law by permitting the jury to consider irrelevant evidence and by lightening the prosecution’s burden of proof. I find neither argument sufficiently persuasive to justify use of the federal test, at least until this court receives more explicit guidance from the United States Supreme Court.
Whether an instruction permitting jury consideration of irrelevant evidence would violate a criminal defendant’s right of due process under the federal Constitution is at present unclear. (Compare Estelle v. McGuire, supra, 502 U.S. _ [116 L.Ed.2d 385, 397-398, 112 S.Ct. 475, 481] [“. . . we need not explore further the apparent assumption of the Court of Appeals that it is a violation of the due process guaranteed by the Fourteenth Amendment for evidence that is not relevant to be received in a criminal trial”], with Barnes v. United States (1973) 412 U.S. 837, 844-845 [37 L.Ed.2d 380, 386-387, 93 S.Ct. 2357] [“Common-law inferences, like their statutory counterparts, must satisfy due process standards in light of present-day experience.”]; see also People v. Castro (1985) 38 Cal.3d 301, 313 [211 Cal.Rptr. 719, 696 P.2d 111] (plur. opn.).) For present purposes, I will assume, without deciding, that the federal Constitution prohibits any state *212rule that would permit a criminal conviction to be based in whole or in part on irrelevant considerations.
Even so, defendant’s argument is defective because propensity evidence is not, as defendant maintains, irrelevant to guilt in a criminal prosecution. Although courts and commentators have given various explanations for the rule excluding propensity evidence, they agree that the evidence has probative value to establish guilt. (See, e.g., Michelson v. United States, (1948) 335 U.S. 469, 476 [93 L.Ed. 168, 174, 69 S.Ct. 213]; People v. Smallwood (1986) 42 Cal.3d 415, 428 [228 Cal.Rptr. 913, 722 P.2d 197]; People v. Alcala, supra, 36 Cal.3d 604, 630-631; People v. Thompson (1980).27 Cal.3d 303, 317 [165 Cal.Rptr. 289, 611 P.2d 883]; People v. Schader (1969) 71 Cal.2d 761, 772 [80 Cal.Rptr. 1, 457 P.2d 841]; IA Wigmore on Evidence (Tillers rev. ed. 1983) § 58.1, p. 1211; 1 McCormick on Evidence (4th ed. 1992) § 190, pp. 797-798.) Because propensity evidence is relevant to guilt, its consideration by the jury cannot violate the federal Constitution’s implied proscription (assuming there is one) against jury consideration of irrelevant evidence.
The decisions of the United States Supreme Court do not establish that errors of the kind at issue here—that is, violations of state rules barring jury consideration of evidence that is relevant but unduly prejudicial—contravene the federal Constitution by lightening the prosecution’s burden of proof. It is true of course that any trial error that has any potential for prejudicing the defendant makes conviction marginally more likely and in this sense lightens the prosecution’s burden of proof. But it cannot be that every such error thereby becomes one of federal constitutional significance. On the contrary, the United States Supreme Court has, in its own words, “ ‘defined the category of infractions that violate “fundamental fairness” very narrowly.’ ” (Estelle v. McGuire, supra, 502 U.S. _[116 L.Ed.2d 385, 398-399, 112 S.Ct. 475, 482], quoting Dowling v. United States (1990) 493 U.S. 342, 352 [107 L.Ed.2d 708, 720, 110 S.Ct. 668].)
The error here did not influence the jury’s understanding of either the elements of the charged offenses or the prosecution’s burden of proof. Therefore, I am not persuaded that the error lightened the prosecution’s burden of proof in a manner constituting a federal constitutional violation.
Having concluded that the argument for federal constitutional error is insubstantial, I would continue to apply our state harmless error standard, under which a judgment will be reversed for error only when, after an examination of the entire cause, including the evidence, the reviewing court is of the opinion that it is reasonably probable that a result more favorable to *213the defendant would have been reached in the absence of the error. (People v. Watson, supra, 46 Cal.2d 818, 836.) Applying this standard here, I find the error harmless.
The majority reasons that the error was not prejudicial because the prosecution presented “overwhelming evidence establishing defendant’s guilt of the charged offenses[.]” (Maj. opn., ante, at p. 187.) But, as defendant points out, the strength of the prosecution’s evidence in this case depended upon the credibility of the prosecution’s witnesses who testified to defendant’s confessions and admissions. Each of these witnesses participated to some extent in the drug operation, and exposure of the drug operation thus threatened each of them, just as it threatened defendant. The charged murders occurred in the apartment of one of these witnesses (Patricia Shepard) and the victims’ bodies were buried at the residence of another witness (Susan Rambo). A third witness (Larry Tom Whittington) admittedly participated in moving the bodies and concealing evidence of the murders. None of the witnesses came forward until long after the murders had occurred, and all had ample time to prepare and coordinate a story placing the blame on defendant. Evidence that depends on the credibility of such witnesses can hardly be deemed overwhelming.
Nevertheless, I am satisfied that defendant was not prejudiced by the court’s instruction permitting the jury to consider the Rambo murder as propensity evidence. The instruction could prejudice defendant only if the jury first determined that it was defendant who had killed Rambo, then inferred from this killing that defendant was a violent person capable of murder, and finally used this inference about defendant’s character to conclude that defendant committed the charged offenses. But the evidence identifying defendant as Rambo’s killer, like the evidence identifying defendant as the perpetrator of the charged murders, consisted primarily of the testimony of other members of the drug ring, with little or no physical evidence tying defendant to the crime. To conclude that defendant had killed Rambo, the jury would have had to find the prosecution’s witnesses credible. If the jury believed them as to Rambo’s murder, it would almost inevitably believe them as to the charged murders as well.
This analysis is consistent with the defense strategy. The defense recognized that the credibility of the prosecution’s witnesses was the key issue— virtually the only issue—in the case. The defense did not oppose admission of the Rambo murder evidence because it viewed that evidence as less credible than the evidence of the charged murders. Thus, the defense was counting on the jury to disbelieve the prosecution’s witnesses as to the Rambo murder, and by this means to undermine their credibility as to the charged murders.
*214Given the nature of the credibility contest presented to the jury, it is highly unlikely that the jury in reaching its guilt verdicts relied to any: significant degree on inferences about defendant’s character drawn from the Rambo murder. For this reason, it is not reasonably probable that the jury would have returned a verdict more favorable to defendant had the trial court not given the erroneous instruction.
Appellant’s petition for a rehearing was denied January 26, 1994.