Opinion
We granted review in this case to examine and clarify the so-called Mendez-Arline rule, which restricts the admission of defense evidence *829tending to show that a third party is guilty of the offense charged. (People v. Mendez (1924) 193 Cal. 39 [223 P. 65]; People v. Arline (1970) 13 Cal.App.3d 200 [91 Cal.Rptr. 520].)
Defendant was convicted of first degree murder and second degree robbery and burglary, and the Court of Appeal affirmed the judgment. After reviewing the relevant law and the record, we also affirm, holding that the error in excluding the proffered evidence was harmless. At the same time, we correct any misapprehension generated by the case law that the Mendez-Arline rule is a principle of exclusion requiring a preliminary showing of a “substantial probability” of third-party guilt before such evidence may be admitted. To the contrary, we reaffirm the admissibility of any relevant evidence that raises a reasonable doubt as to a defendant’s guilt, including evidence tending to show that a party other than the defendant committed the offense charged. Such evidence may be excluded only when the court properly exercises its discretion under Evidence Code section 352 to reject evidence that creates a substantial danger of undue consumption of time or of prejudicing, confusing, or misleading the jury.1
The body of Israel Deasonhouse was found on the kitchen floor of his home on December 31, 1981. He was lying face-up, his clothing in disarray, his pants pockets fully turned out. He apparently had been eating cottage cheese at the time of his death, and had defecated in his pants. The investigator noted a distinctive hemorrhaging under the victim’s eyelids that may indicate death by air deprivation, but finding no external signs of trauma and no evidence of forced entry he ordered a noncriminal autopsy to determine cause of death. The coroner determined the cause of death was cardiac arrhythmia incident to heart disease. The examination was limited to the victim’s heart and chest.
In October 1982, on the basis of information volunteered by one Rhae Foust, the authorities ordered the body exhumed and a criminal autopsy performed. Investigation of the neck area revealed a fractured hyoid bone and local hemorrhaging, suggesting that the fatal cardiac arrythmia may have been caused by strangulation.
Foust had come forward with information after his arrest for drunk driving, apparently hoping to mitigate punishment for that offense and for forgeries he had allegedly committed with defendant, although the police clearly advised him that no charges would be dropped in exchange. He recounted that on December 31, 1981, defendant told him that he and another friend, David Rodriguez, had killed an elderly man named Israel. He stated that *830defendant described the victim as eating cottage cheese and as losing bowel control when he was strangled alternately by defendant and Rodriguez, and also told of returning to the victim’s house to confirm that he was dead. In a later conversation, Foust continued, defendant explained that he had driven the victim to the bank the day of killing and had seen what appeared to be a large roll of money, but that he had found only a small sum in his pockets during the murder. Foust attributed his failure to come forward earlier to his fear of reprisal by Rodriguez or defendant, who had threatened to kill him if he told anyone.
Defendant was arrested. He admitted in a statement to the police that he and Rodriguez gave Deasonhouse a ride to the bank, but denied participating in the murder. He claimed Rodriguez was the killer. Rodriguez, he said, told him that Deasonhouse answered his knock at the door, and that Rodriguez immediately seized and choked him. Defendant said nothing about the cottage cheese or the involuntary defecation. He conceded, however, that he might have boasted to Foust of participating in the murder in order to impress him.
The defense attempted to show that Deasonhouse was killed by someone other than defendant. After the prosecution rested, defendant made an offer of proof to show that Deasonhouse was more likely murdered by a teenage neighbor who had been seen victimizing him in the past, or by Foust himself. Defendant intended to base his theory of Foust’s culpability on the latter’s knowledge of intimate details of the murder not mentioned by defendant; on the fact that the victim’s hyoid bone was broken on the left side, although defendant is right-handed; and on the proposed testimony of Foust’s estranged wife that he is left-handed, is violent when drunk, and has a history as a police informant. She would further testify that Foust had never spoken of a meeting with defendant or expressed fear of him, and was “virtually tantamount to a pathological liar.” Defendant would also attempt to prove that Foust’s motive for bringing the murder to the attention of police was to avoid prosecution for forgery.
The court responded that “the test ... as I understand it from reading [People v. Arline, supra, 13 Cal.App.3d 200] is whether the proffered evidence would show more than a possible ground or possible suspicion that Rhae Foust committed the offenses charged in the Information.” After reviewing the proposed testimony, the court reasoned, “I don’t see how your evidence, as I say, gets beyond the point of possible suspicion, since [it is] almost incredible] for a person to make statements concerning a murder that he himself committed ten months after the murder when he must have reasonably been aware of the fact that law enforcement probably had, to some degree, given up on it.” The court concluded by reciting the words *831of section 352 but determined only that the evidence lacked probative value under the Arline test. It therefore denied the offer of proof.
Nevertheless, the defense thereafter succeeded in introducing a significant amount of the evidence it believed linked Foust to the crime. Foust’s wife testified that her husband never spoke of Deasonhouse’s murder or mentioned fearing defendant, and that he was generally untruthful. Another witness testified that Foust was violent; she also said he wore “waffle-stomper” shoes around the time of the murder, to impeach testimony by Foust that he never wore such footwear. Defendant’s closing argument emphasized to the jury that “waffle-stomper” prints were found in Deasonhouse’s bedroom.
Defendant further argued that Foust was left-handed and the victim’s hyoid bone was broken on the left side, a fact suggesting a left-handed perpetrator. He stressed that Foust revealed details to the police that only someone present at the murder could know, while defendant related a more sparse version and denied any knowledge of what the victim had been eating or his loss of bowel control. Finally, he suggested Foust had a motive to offer information to the authorities—to avoid prosecution for drunk driving and forgery by currying favor with the police.
I
Defendant contends that the court denied him the right to present more than a minimal defense founded on evidence that came in “through the back door,” and that the court’s reliance on Arline was error. We agree. What has come to be called the Mendez-Arline test is an exclusionary rule abstracted from case law articulating the minimum threshold for admitting evidence of third-party culpability that tends to exonerate a defendant. Simply expressed, it provides that such evidence is admissible only if it constitutes “substantial evidence tending to directly connect that person with the actual commission of the offense.” (People v. Green (1980) 27 Cal.3d 1, 22 [164 Cal.Rptr. 1, 609 P.2d 468]; see generally Garcia, Third Party Culpability Evidence: A Criticism of the California Mendez-Arline Exclusionary Rule—Towards a Constitutional Standard of Relevance and Admissibility (1983) 17 U.S.F. L.Rev. 441.)
The rule takes its name from two key formulations of the standard for admitting evidence of third-party culpability: People v. Mendez, supra, 193 Cal. 39,2 the first California Supreme Court case upholding the exclusion *832of such evidence, and People v. Arline, supra, 13 Cal.App.3d 200, the first Court of Appeal opinion to restate that standard after enactment of section 352. At the outset, therefore, it is important to recall both the specific articulations of the rule in those cases as well as the factual contexts in which they emerged.
The defendant in Mendez offered to prove that four Mexican workers residing a quarter of a mile from the scene of a ranchowner’s murder had quarrelled with the victim, and that their disappearance after quitting employment on the morning after the homicide suggested flight and thereby admission of guilt. Witnesses had testified they had seen three Mexican men in the vicinity of the crime on the day it occurred. The trial court rejected testimony about the victim’s dispute with the missing laborers, concluding that such evidence of possible motive was not linked to any other evidence “having the inherent tendency” to connect them to the commission of the crime. It similarly rejected testimony describing three of the four men who disappeared because, as the defendant conceded, the descriptions were “so general and indefinite that they would apply equally to hundreds of persons.” (193 Cal. at p. 51.)
This court affirmed. We noted that it is always proper to defend against criminal charges by showing that a third person, and not the defendant, committed the crime charged. On the other hand, we held although a defendant should not be required to establish the guilt of a third person “with that degree of certainty requisite to sustain a conviction of the latter,” exculpatory evidence pointing to that person should not be admitted if it “simply affords a possible ground of possible suspicion. ...” (Ibid.) Similarly, “mere evidence of motive in another person, or of motive coupled with threats of such other person, is inadmissible unless coupled with other evidence tending to directly connect such other person with the actual commission of the crime charged.” (Ibid.) The court observed that the rule rests on the sound basis that evidence must be both relevant and material. In the interests of judicial order and expediency, strict limits must be imposed on the scope of inquiry into collateral or insignificant matters that might suggest third-party culpability. (Id. at p. 52.)
In People v. Arline, supra, 13 Cal.App.3d 200, the Court of Appeal rearticulated those limits. To be admissible, it declared, evidence implicating a third party in lieu of the defendant must rise to the level of “‘substantial proof of a probability that this happened.’ ” (Id. at p. 204.) Charged with robbing a service station, the defendant in Arline offered to show that two months after the incident another person of similar appearance robbed a different service station using a similar modus operand!. He argued that *833the perpetrator was in the area at the time of the first incident, and could therefore have been responsible. The trial court excluded the evidence, finding the two men did not resemble each other and the ways in which the crimes were committed differed both as to weapons used and number of participants.
Citing both section 352 and precode case law including Mendez, the Court of Appeal sustained the ruling: the proffered evidence was of “no probative value and little if any relevance,” yet was likely to generate “suspicion and, hence, prejudice in the minds of jurors.” (Id. at p. 205.) The court noted that innumerable cases support the exclusion of even relevant evidence that is unduly prejudicial to the defendant, and suggested that the prosecution likewise “is entitled to the protection of the court from prejudicial evidence.” (Ibid.)
As expressed in the Arline opinion, the basis for excluding evidence of third-party guilt is ambiguous. On the one hand, the opinion purports simply to apply section 352 (ibid.); on the other, there is language clearly calling for a threshold showing of “ ‘substantial proof of a probability’ ” of such guilt. (Id. at p. 204.) We believe the latter language sets too high a standard. As Justice Poché noted in his criticism of Arline, it improperly “tends to exclude evidence of third party culpability on all but extraordinarily strong showings.” (People v. Perry (1980) 104 Cal.App.3d 268, 276 [163 Cal.Rptr. 522] (dis. opn. of Poché, J.)
To be admissible, the third-party evidence need not show “substantial proof of a probability” that the third person committed the act; it need only be capable of raising a reasonable doubt of defendant’s guilt. At the same time, we do not require that any evidence, however remote, must be admitted to show a third party’s possible culpability. As this court observed in Mendez, evidence of mere motive or opportunity to commit the crime in another person, without more, will not suffice to raise a reasonable doubt about a defendant’s guilt: there must be direct or circumstantial evidence linking the third person to the actual perpetration of the crime.
In the present case, defendant submitted offers of proof linking Foust to the actual murder: waffle-stomper prints in the victim’s bedroom, the likely left-handedness of the killer, and knowledge of unique particulars of the murder all pointed to Foust as the possible killer. On the basis of Arline, however, the trial court concluded that defendant was required to make a preliminary showing of “substantial probability” that the third party actually committed the crime. This finding was error. The court’s proper inquiry was limited to whether this evidence could raise a reasonable doubt as to defendant’s guilt and then applying section 352.
*834We reject Arline to the extent that it creates a distinct and elevated standard for admitting this kind of exculpatory evidence.3 Rather than speaking in terms of a Mendez-Arline “rule,” courts should simply treat third-party culpability evidence like any other evidence: if relevant it is admissible (§ 350) unless its probative value is substantially outweighed by the risk of undue delay, prejudice, or confusion (§ 352). We recognize that an inquiry into the admissibility of such evidence and the balancing required under section 352 will always turn on the facts of the case. Yet courts must weigh those facts carefully. They should avoid a hasty conclusion such as the trial court’s finding in the present case that evidence of Foust’s guilt was “incredible.” Such a determination is properly the province of the jury.
Furthermore, courts must focus on the actual degree of risk that the admission of relevant evidence may result in undue delay, prejudice, or confusion. As Wigmore observed, “if the evidence is really of no appreciable value no harm is done in admitting it; but if the evidence is in truth calculated to cause the jury to doubt, the court should not attempt to decide for the jury that this doubt is purely speculative and fantastic but should afford the accused every opportunity to create that doubt.” (1A Wigmore, Evidence (Tillers rev. ed. 1980) § 139, p. 1724.)
II
Defendant contends his constitutional right to present a defense precludes any application whatever of section 352 to third-party culpability evidence: even remote evidence of motive without more, he argues, might raise a “reasonable doubt” of guilt. Indeed, defendant insists, the concept of reasonable doubt is itself so elusive that any attempt to weigh the probative value of such evidence must fail. He asserts that no amount of time spent presenting such a defense could be regarded as “undue,” and the only prejudice the court must avert is “emotional bias.”
The claim does not withstand scrutiny. As a general matter, the ordinary rules of evidence do not impermissibly infringe on the accused’s right to present a defense. Courts retain, moreover, a traditional and intrinsic power to exercise discretion to control the admission of evidence in the interests of orderly procedure and the avoidance of prejudice. (People v. Castro (1985) 38 Cal.3d 301, 306-307 [211 Cal.Rptr. 719, 696 P.2d 111]; People v. Reeder (1978) 82 Cal.App.3d 543, 552 [147 Cal.Rptr. 275].) As we observed in Mendez, this principle applies perforce to evidence of third-*835party culpability: “if evidence of motive alone upon the part of other persons were admissible, ... in a case involving the killing of a man who had led an active and aggressive life it might easily be possible for the defendant to produce evidence tending to show that hundreds of other persons had some motive or animus against the deceased . . . .” (People v. Mendez, supra, 193 Cal. at p. 52.) Trials must reach an end, and that end must be logical. (Ibid.)
Although defendant cites a wealth of out-of-state cases ostensibly supporting his constitutional argument, not one suggests that third-party culpability evidence must be admitted no matter how remotely relevant and without considering its potential to disrupt orderly proceedings. Likewise, however liberal the federal courts had been under Alexander v. United States (1891) 138 U.S. 353 [34 L.Ed. 954, 11 S.Ct. 350]-which excludes only third-party culpability evidence having “no legitimate tendency” to exculpate a defendant—since the advent of the Federal Rules of Evidence all such evidence is subject to screening under federal rule 403 (28 U.S.C.), the counterpart of our section 352. Defendant’s alternate proposition, that the prosecution be required to show a compelling state interest before evidence of third-party culpability may be excluded, similarly overlooks the policy behind section 352, which incorporates a compelling state interest, i.e., the policy in favor of an orderly trial on the merits.
Ill
Although we hold that the court improperly rejected defendant’s offer of proof of Rhae Foust’s culpability, we conclude that on this record the error was harmless. Despite the ruling, the only evidence that would potentially link Foust to the crime was adequately placed before the jury: his knowledge of intimate details of the murder, the presence of “wafflestomper” prints in the victim’s home, and his left-handedness. In closing argument, after putting all these circumstantial indications of Foust’s possible guilt together, the defense clearly and repeatedly proposed that Foust, not defendant, committed the murder.
Equally important, defendant’s theory of Foust’s culpability would not tend to exculpate him in any event. Because no testimony or circumstantial evidence limited the number of perpetrators, Foust’s participation would not undermine the significant evidence linking defendant to the murder. Defendant himself admitted driving Deasonhouse to the bank and discussing with Rodriguez robbing Deasonhouse of his roll of bills; he admitted boasting to Foust of killing “an old man named Israel” before notice of Deasonhouse’s death appeared in the newspaper; and his own alibi placed him, *836together with Rodriguez, within two blocks of Deasonhouse’s home on the probable night of the murder, on a joint venture to purchase amphetamines.
Defendant’s only evidence that would have clearly implicated Foust was that the latter was lying—not about defendant’s confession shortly after the murder, which he admits—but about learning specific details of the crime from defendant. Yet two witnesses gave testimony impeaching Foust’s veracity, including his wife who accused him of habitual lying and untrustworthiness.
The sole proposed evidence that was not otherwise before the jury was defendant’s claim that Foust had a heavy drug habit and a history of informing for the police. Again that evidence does not tend to exculpate defendant. The jury did hear testimony that Foust, together with defendant, injected narcotics and became intoxicated on occasion. Defendant did not offer evidence of a serious addiction to drugs; but even had he done so, it would have been so remote as to be of little if any relevance as evidence of motive. Finally, evidence of Foust’s alleged past cooperation with police is simply immaterial. In these circumstances, we conclude it is not reasonably probable that a result more favorable to defendant would have been reached in the absence of the error. (People v. Watson (1956) 46 Cal.2d 818, 837 [299 P.2d 243].)
The judgment of the Court of Appeal is affirmed.
Bird, C. J., Broussard, J., Reynoso, J., Grodin, J., Lucas, J., and Delucchi (Alfred A.), J.,* concurred.