Petitioner John George Brown was convicted of capital murder, the underlying facts of which are fully recounted in People v. Brown (1988) 46 Cal.3d 432 [250 Cal.Rptr. 604, 758 P.2d 1135] (Brown I). Briefly, petitioner shot and killed Garden Grove Police Officer Donald Reed. Reed and several other officers were pursuing him as he was leaving the Cripple Creek Bar. (Id. at p. 440.) Upon reaching the exit, petitioner turned and fired eight shots, killing Reed and seriously wounding two other officers and two civilians. Once outside the bar, he hid in some bushes where he was located two hours later. The probable murder weapon, a .22-caliber semiautomatic handgun, was found nearby. At trial, petitioner asserted a defense of diminished capacity, claiming he had been under the influence of methamphetamine. (Id. at p. 441; see former Pen. Code, § 22.) The prosecution rebutted this claim with evidence a sample of his blood had tested negative for any drugs. (Brown I, supra, 46 Cal.3d at p. 441.)
A jury convicted petitioner of first degree murder and found true the special circumstance allegation of intentionally killing a peace officer engaged in the performance of his duties. (Pen. Code, § 190.2, subd. (a)(7).) Following penalty phase evidence, the jury determined the punishment should be death. (Brown I, supra, 46 Cal.3d at pp. 441-442.) On automatic appeal, this court affirmed the judgment.
*877Thereafter, petitioner sought a writ of habeas corpus, alleging the prosecution committed Brady error1 in failing to disclose that the result of a radioactive immunoassay (RIA) of his blood sample had been positive for phencyclidine (PCP). The negative results introduced at trial were based on gas chromatography mass spectrometry (GC/MS) testing.
We issued an order to show cause. Because of factual conflicts raised by respondent’s return, we ordered a reference hearing to resolve the following questions:
Did the prosecution disclose the positive PCP finding to petitioner, his investigator, or his counsel before or during trial?
Did the positive PCP finding from the RIA test indicate that there was PCP or a PCP analog in petitioner’s blood at the time of the crimes? If so, did the subsequent negative PCP finding from the GC/MS test establish that there was no PCP or PCP analog in petitioner’s blood at the time of the crimes? How can the results of the two tests be reconciled?
Having considered the record of the hearing and petitioner’s original trial in light of controlling United States Supreme Court authority, we conclude the prosecution failed to disclose material exculpatory evidence and, therefore, now grant relief.
1. Factual findings
On the disclosure question, the evidence at the reference hearing established that the Orange County Sheriff-Coroner, Department of Forensic Science Services (the crime lab), which tested petitioner’s blood, utilized a multipage form to record the results of its toxicological examination. The top page, on which lab personnel recorded the GC/MS result, was referred to as the “result sheet.” The bottom page, with the RIA result, was designated the “worksheet.” At the time petitioner’s blood was tested, the standard procedure of the crime lab upon completion of the toxicology work was to send a copy of the result sheet to the prosecutor and defense counsel; for reasons of laboratory protocol, however, a copy of the worksheet was sent only on specific request. According to Frank Fitzpatrick, the chief criminalist responsible for management of the crime lab clerical staff at the time, notations on petitioner’s worksheet indicated to him a copy had been sent to *878Deputy Public Defender Michael Beecher on October 16, 1980, via county messenger.
Michael Beecher, who represented petitioner from the preliminary hearing through November 1980 when he became advisory counsel, testified that on October 16, 1980, he telephoned the crime lab to request information on petitioner’s blood testing and received a copy of the result sheet the next day. He had no recollection, file notation, or other indication he ever requested or received a copy of the worksheet. Petitioner’s trial counsel, Daye Shinn, also testified he did not receive or see a copy of the worksheet before or during trial, nor did petitioner or his investigator.2 The prosecutor also was unaware of the document until it was brought to his attention in conjunction with the present proceedings. The referee examined the case files of the public defender, the district attorney, and the Garden Grove Police Department and found no copy of the worksheet in any of them.
With respect to our first question, the referee found the witnesses on both sides credible, but ultimately concluded “the evidence preponderates that the documentation, including the positive RIA finding, was freely available, disclosed and duly forwarded to defense counsel before trial, as requested.” He also determined there was no evidence “any police agency or the District Attorney withheld or concealed information.” Petitioner objects to these and related findings as unsupported by the record. We need not definitively resolve his objections for the more fundamental reason that these findings do not accurately respond to the specific question posed in our reference order—did the prosecution disclose the RIA result to petitioner or anyone acting on his behalf? Relevant to this query, the referee impliedly found the prosecutor had no knowledge of the worksheet and thus could not have disclosed it. The referee also drew “a reasonable inference that [the worksheet] was not received” by Beecher based on testimony he was unaware of its existence. The uncontradicted evidence also confirms neither petitioner nor Shinn ever saw the document, and the referee makes no findings, express or implied, to the contrary.
As we explain, these latter findings are determinative. Responsibility for Brady compliance lies exclusively with the prosecution, including the “duty to learn of any favorable evidence known to the others acting on the government’s behalf in the case.” (Kyles v. Whitley (1995) 514 U.S. 419, 437 [115 S.Ct. 1555, 1567, 131 L.Ed.2d 490] (Kyles).) Whatever the reason for failing to discharge that obligation, the prosecution remains accountable for the consequence. (Id. at pp. 437-438 [115 S.Ct. at pp. 1567-1568].)
*8792. Governing legal principles
Pursuant to Brady, supra, 373 U.S. 83, the prosecution must disclose material exculpatory evidence whether the defendant makes a specific request (id. at p. 87 [83 S.Ct. at pp. 1196-1197]), a general request, or none at all (United States v. Agurs (1976) 427 U.S. 97, 107 [96 S.Ct. 2392, 2399, 49 L.Ed.2d 342] (Agurs)). The scope of this disclosure obligation extends beyond the contents of the prosecutor’s case file and encompasses the duty to ascertain as well as divulge “any favorable evidence known to the others acting on the government’s behalf . . . .” (Kyles, supra, 514 U.S. at p. 437 [115 S.Ct. at p. 1567].) Courts have thus consistently “decline[d] ‘to draw a distinction between different agencies under the same government, focusing instead upon the “prosecution team” which includes both investigative and prosecutorial personnel.’ ” (United States v. Auten (5th Cir. 1980) 632 F.2d 478, 481.)3 “A contrary holding would enable the prosecutor ‘to avoid disclosure of evidence by the simple expedient of leaving relevant evidence to repose in the hands of another agency while utilizing his access to it in preparing his case for trial,’ [citation].” (Martinez v. Wainwright, supra, 621 F.2d at p. 188; United States ex rel. Smith v. Fairman (7th Cir. 1985) 769 F.2d 386, 391-392.) Thus, “whether the nondisclosure was a result of negligence or design, it is the responsibility of the prosecutor. The prosecutor’s office is an entity and as such it is the spokesman for the Government.” (Giglio v. United States (1972) 405 U.S. 150, 154 [92 S.Ct. 763, 766, 31 L.Ed.2d 104]; Kyles, supra, 514 U.S. at p. 439 [115 S.Ct. at p. 1568].)
As a concomitant of this duty, any favorable evidence known to the others acting on the government’s behalf is imputed to the prosecution. “The individual prosecutor is presumed to have knowledge of all information gathered in connection with the government’s investigation.” (U.S. v. Payne (2d Cir. 1995) 63 F.3d 1200, 1208 (Payne); see Smith v. Secretary Dept. of Corrections, supra, 50 F.3d at pp. 824-825, and cases cited therein.) The Supreme Court recently reiterated this principle: “whether the prosecutor succeeds or fails in meeting this obligation [to learn of favorable evidence] *880(whether, that is, a failure to disclose is in good faith or bad faith, [citation]), the prosecution’s responsibility for failing to disclose known, favorable evidence rising to a material level of importance is inescapable.” (Kyles, supra, 514 U.S. at pp. 437-438 [115 S.Ct. at pp. 1567-1568]; see also Giglio v. United States, supra, 405 U.S. at p. 154 [92 S.Ct. at p. 766].)4
3. Application to the facts
In this case, the uncontroverted record confirms that neither petitioner nor anyone acting on his behalf received a copy of the worksheet. At the reference hearing, Fitzpatrick acknowledged the lab “worked closely with the District Attorney’s Office in assisting it in the prosecution of cases”; and there is no serious dispute that in these circumstances it was part of the investigative “team.” The prosecutor thus had the obligation to determine if the lab’s files contained any exculpatory evidence, such as the worksheet, and disclose it to petitioner. Whether or not he actually did examine the files, the lab personnel’s knowledge is imputed.
In United States ex rel. Smith v. Fairman, supra, 769 F.2d 386, an investigating officer examined a gun the defendant allegedly used to shoot at two police officers and commit a robbery. The day after the shooting, the investigator recorded on a firearms worksheet that he found the gun inoperable. (Id. at p. 389.) When he prepared his official report, however, he failed to include this information “because the only bullets found at the scene had been fired from [one of the officers’] gun[s]”; “it was departmental procedure only to report the results of a match between the bullets recovered from the scene of a crime and the guns submitted for testing.” (Id. at pp. 389-390, fn. omitted.) The firearms report was placed in the investigator’s files and not disclosed to the prosecutor or the defendant. Although the prosecutor had no actual knowledge of the information, the reviewing court found Brady error in light of the “closely aligned” working relationship between the investigating officer and the prosecution. (Id. at p. 391.) “We believe that the purposes of Brady would not be served by allowing material exculpatory evidence to be withheld simply because the police, rather than the prosecutors, are responsible for the nondisclosure. [Citation.] . . . The fact that the prosecutor in this case was blameless therefore does not justify the State’s failure to produce [the] firearms worksheet.” (Id. at pp. 391-392.)
Any other rule would leave the defendant’s due process rights to the fortuity of a subordinate agency’s procedural protocol, which the Supreme *881Court has squarely rejected. “[A]ny argument for excusing a prosecutor from disclosing what he does not happen to know about boils down to a plea to substitute the police for the prosecutor, and even for the courts themselves, as the final arbiters of the government’s obligation to ensure fair trials.” (Kyles, supra, 514 U.S. at p. 438 [115 S.Ct. at p. 1568].) Similarly in this case, the crime lab’s failure to apprise the prosecution of the worksheet did not relieve the prosecutor of his obligation to review the lab’s files for exculpatory evidence. Thus, he remained ultimately responsible when the defense did not receive a copy.
Nor would the crime lab’s attempt to transmit the worksheet be sufficient to satisfy the prosecutor’s obligation. Such a result would be fundamentally at odds with the due process imperatives of the disclosure rule. The principles Brady and its progeny embody are not abstractions or matters of technical compliance. (Cf. United States v. Leon (1984) 468 U.S. 897 [104 S.Ct. 3405, 82 L.Ed.2d 677].) The sole purpose is to ensure the defendant has all available exculpatory evidence to mount a defense. To that end, a document sent but not received is as useless as a document not sent at all. In both situations, the right to a fair trial is equally denied. A comparable situation arose in United States v. Consolidated Laundries Corporation (2d Cir. 1961) 291 F.2d 563, in which, “[t]hrough unexplained acts, certain of the [material exculpatory documents] in the Government’s custody were misplaced and consequently were not shown to defendants when inspection was ordered. This failure to produce deprived the defendants of evidence which would have been helpful to them .... For practical purposes the [documents] were temporarily lost. We hold this was negligence chargeable to the prosecution.” (Id. at p. 570.)
Equally important, the Supreme Court has unambiguously assigned the duty to disclose solely and exclusively to the prosecution; those assisting the government’s case are no more than its agents. (Kyles, supra, 514 U.S. at p. 438 [115 S.Ct. at p. 1568]; Giglio v. United States, supra, 405 U.S. at p. 154 [92 S.Ct. at p. 766]; Fero v. Kerby (10th Cir. 1994) 39 F.3d 1462, 1472, fn. 12.) By necessary implication, the duty is nondelegable at least to the extent the prosecution remains responsible for any lapse in compliance. Since the prosecution must bear the consequences of its own failure to disclose (see, e.g., U.S. v. Ellis (4th Cir. 1997) 121 F.3d 908, 914 (Ellis); United States v. Consolidated Laundries Corporation, supra, 291 F.2d at p. 570), a fortiori, it must be charged with any negligence on the part of other agencies acting in its behalf (Fero v. Kerby, supra, 39 F.3d at p. 1472, fn. 12; cf. Ellis, supra, 121 F.3d at p. 914 [defense counsel’s failure to renew request for witness statements at trial does not discharge prosecution’s Brady obligation]; U.S. v. Alvarez (9th Cir. 1996) 86 F.3d 901, 905 (Alvarez) [delegating to nonattorney police officer responsibility to determine if officers’ rough notes contain *882Brady material deemed “problematic”]; Walker v. City of New York (2d Cir. 1992) 974 F.2d 293, 299 [“It is appropriate that the prosecutors, who possess the requisite legal acumen, be charged with the task of determining which evidence constitutes Brady material that must be disclosed to the defense. A rule requiring the police to make separate, often difficult, and perhaps conflicting, disclosure decisions would create unnecessary confusion.”]). Accordingly, the risk and consequences of nonreceipt must fall to the prosecution.
Despite any seeming unfairness to the prosecution, no other result would satisfy due process in this context. “The principle ... is not punishment of society for misdeeds of a prosecutor but avoidance of an unfair trial to the accused.” (Brady, supra, 373 U.S. at p. 87 [83 S.Ct. at p. 1197]; Kyles, supra, 514 U.S. at pp. 437-438 [115 S.Ct. at pp. 1567-1568]; Smith v. Phillips (1982) 455 U.S. 209, 219 [102 S.Ct. 940, 947, 71 L.Ed.2d 78]; see Agurs, supra, 427 U.S. at p. 110 [96 S.Ct. at p. 2401] [Constitutional error under Brady results from “the character of the evidence, not the character of the prosecutor.”].) Declining to hold the state “accountable under Bagley and Brady for evidence known only to police investigators and not to the prosecutor . . . would . . . amount to a serious change of course from the Brady line of cases.” (Kyles, supra, 514 U.S. at p. 438 [115 S.Ct. at p. 1568], fn. omitted.) “[T]he prosecutor has ready access to a veritable storehouse of relevant facts and, within the ambit of constitutional, statutory and jurisprudential directives, this access must be shared ‘in the interests of inherent fairness ... to promote the fair administration of justice.’ [Citation.] If disclosure were excused in instances where the prosecution has not sought out information readily available to it, we would be inviting and placing a premium on conduct unworthy of representatives of the [government].” (United States v. Auten, supra, 632 F.2d at p. 481.) As one court has expressed it: “We suspect the courts’ willingness to insist on an affirmative duty of inquiry may stem primarily from a sense that an inaccurate conviction based on government failure to turn over an easily turned rock is essentially as offensive as one based on government non-disclosure. [Citation.]” (Brooks, supra, 966 F.2d at p. 1503.)
“In the State’s favor it may be said that no one doubts that police investigators sometimes fail to inform a prosecutor of all they know. But neither is there any serious doubt that ‘procedures and regulations can be established to carry [the prosecutor’s] burden and to insure communication of all relevant information on each case to every lawyer who deals with it.’ [Citation.]” (Kyles, supra, 514 U.S. at p. 438 [115 S.Ct. at p. 1568]; see, e.g., United States ex rel. Smith v. Fairman, supra, 769 F.2d at p. 390 & fn. 1; Carey v. Duckworth, supra, 738 F.2d at p. 878 & fn. 4.) Thus, while the *883Constitution does not impose a duty “to allow complete discovery of [the prosecutor’s] files as a matter of routine practice” (Agurs, supra, 427 U.S. at p. 109 [96 S.Ct. at p. 2400]), the Supreme Court has on more than one occasion urged “the careful prosecutor” to err on the side of disclosure and, by necessary extension, thorough examination of investigative files. (Kyles, supra, 514 U.S..at p. 440 [115 S.Ct. at p. 1569]; Agurs, supra, 427 U.S. at p. 108 [96 S.Ct. at pp. 2399-2400]; Giglio v. United States, supra, 405 U.S. at p. 154 [92 S.Ct. at p. 766]; see Alvarez, supra, 86 F.3d at p. 905.)
Although rigorous, we do not perceive the duty imposed by Brady as too onerous. (Cf. In re Littlefield (1993) 5 Cal.4th 122, 135 [19 Cal.Rptr.2d 248, 851 P.2d 42] [cataloguing extent of prosecution’s obligation to disclose in discovery context].) “Obviously some burden is placed on the shoulders of the prosecutor when he is required to be responsible for those persons who are directly assisting him in bringing an accused to justice. But this burden is the essence of due process of law. It is the State that tries a man, and it is the State that must insure that the trial is fair.” (Moore v. Illinois (1972) 408 U.S. 786, 809-810 [92 S.Ct. 2562, 2575, 33 L.Ed.2d 706] (conc. and dis. opn. of Marshall, J.); Alvarez, supra, 86 F.3d at p. 905.)5 This obligation serves “to justify trust in the prosecutor as ‘the representative ... of a sovereignty . . . whose interest . . . in a criminal prosecution is not that it shall win a case, but that justice shall be done.’ ” (Kyles, supra, 514 U.S. at p. 439 [115 S.Ct. at p. 1568], quoting Berger v. United States (1935) 295 U.S. 78, 88 [55 S.Ct. 629, 633, 79 L.Ed. 1314]; United States v. Auten, supra, 632 F.2d at p. 481.) It also tends “to preserve the criminal trial, as distinct from the prosecutor’s private deliberations [or some other agency’s independent assessment of materiality], as the chosen forum for ascertaining the truth about criminal accusations. [Citations.]” (Kyles, supra, 514 U.S. at p. 440 [115 S.Ct. at p. 1568]; Alvarez, supra, 86 F.3d at p. 905.)
Here, as in most circumstances, Brady compliance demanded no more than simple advertence. The evidence was readily accessible to the prosecution. A cursory review of the crime lab’s file would have revealed the existence of the RIA test result. At the reference hearing, the prosecutor acknowledged that he would have promptly divulged the information had he been aware of it. Regardless of the reason, he failed to do so. Under established constitutional principles, we thus resolve the threshold inquiry as to disclosure in petitioner’s favor.
*884B. Materiality of the RIA test
In light of this conclusion, we turn to the question of materiality, for not every nondisclosure of favorable evidence denies due process. “[S]uch suppression of evidence amounts to a constitutional violation only if it deprives the defendant of a fair trial. Consistent with ‘our overriding concern with the justice of the finding of guilt,’ [citation] a constitutional error occurs, and the conviction must be reversed, only if the evidence is material in the sense that its suppression undermines confidence in the outcome of the trial.” (United States v. Bagley (1985) 473 U.S. 667, 678 [105 S.Ct. 3375, 3381, 87 L.Ed.2d 481] (Bagley).)
1. Factual findings
With respect to the differing RIA and GC/MS test results, the evidence at the reference hearing established that the crime lab utilized the RIA as a screening mechanism for PCP and PCP analogs. Because in 1980 the process had certain limitations, scientific protocol required confirmation of a positive result by GC/MS testing. Dr. Vina Spiehler, an expert who testified for respondent, explained that various circumstances, such as cross-reactivity (detection of some other drug) or flaking (incomplete precipitation before final analysis) as well as operator error, could cause the RIA to produce a false positive. In Spiehler’s experience, such a false positive could account for a discrepancy between RIA and GC/MS test results. A disparity might also arise in two other situations: if the concentration of PCP in the test sample were below the cutoff standard for the GC/MS, or if the sample contained a PCP analog. Because the GC/MS process is drug-specific, it would require the use of a different standard to confirm the presence of an analog. Based on her research and experience, Spiehler opined that the probability “a positive RIA actually represented PCP in that sample” was 37 percent.6 Spiehler’s testimony substantially corresponded to that of Dr. Ferris Pitts, petitioner’s expert.
Notes recorded by Bonnie Driver, the criminalist who performed the testing of petitioner’s blood, indicated she used a 25-nanogram-per-milliliter calibration standard when she ran the GC/MS test. According to Fitzpatrick, the cutoff “[c]ertainly [was] far below 25 nanograms. Possibly as low as ten nanograms.” However, in her declaration submitted in support of respondent’s return, Driver states, “At this time I am unable to say if the cutoff level was 10 ng or below.” In Pitts’s opinion, only the graph printouts run on *885the standard and blanks used to calibrate the GC/MS test would establish the cutoff. Those printouts were unavailable and had apparently been destroyed by the crime lab. According to Pitts, the RIA was generally more sensitive than the GC/MS test and could detect PCP “[s]ometimes as low as one or two nanograms, per milliliter in the original blood sample” (the equivalent of 0.000001 gram per liter). Spiehler also testified experts in the field “thought it might be possible to detect as little as . . . two nanograms [of PCP] per milliliter [by RIA]
In response to our series of questions concerning the differing test results, the referee’s findings focused on the RIA’s function as a screening mechanism and the possibility of false positives. He also adopted Spiehler’s probability analysis. The referee concluded the GC/MS result “show[ed] that PCP was not detectable in petitioner’s blood” because printout graphs did not have any “mass points” (numerical indications of the test sample’s chemical structure) at the appropriate ion levels. According to his findings, a mass point would also appear if a PCP analog were present because it “would be expected to have the same ‘backbone’ as PCP.”
The referee reconciled the differing results by essentially discounting the RIA as a false positive due to the “inherent unreliability of the test . . . .” Because the GC/MS test was “more reliable and precise” and its result was “considered final” by the crime lab, the “negative” finding for PCP or PCP analogs was the more accurate. Accordingly, in answer to our predicate questions he found the positive RIA did not indicate there was PCP or an analog in petitioner’s blood at the time of the crimes and the subsequent negative GC/MS established there was not PCP or an analog in his blood.
Petitioner objects to these findings as inaccurate, and we find the objections well taken to the following extent. Nothing in the record establishes the RIA is inherently unreliable. In fact, both experts testified that studies had shown a 98 percent correspondence between RIA and GC/MS testing. Furthermore, since the GC/MS test is in fact more “precise,” i.e., specific, for PCP, it could not conclusively negate the presence of a PCP analog— particularly one that did not share sufficient molecular characteristics with PCP. Spiehler testified some analogs have a “different psychalic structure in the backbone,” which would not show up on a GC/MS run for PCP. Also contrary to the referee’s finding, we find no evidence “blanks and standards” established the GC/MS cutoff level “possibly as low as” 10 nanograms per milliliter. Driver’s declaration did not support this conclusion, and it appears undisputed the printout graphs for the blanks and standards had been lost or destroyed.
While we do not discount the referee’s ultimate findings, they are incomplete with respect to the questions posed in our reference order. The differing results can be reconciled on several bases that do not exclude the *886possibility each test was accurate. If the amount of PCP were below the GC/MS cutoff, the RIA could produce a positive because it is a more sensitive testing procedure. Or, if the substance were an analog that the GC/MS PCP standard would not detect, the RIA would produce a positive result because it is intended to test not only for PCP but for analogs as well, which number over 100. In either situation, the RIA would establish the presence of PCP or an analog in petitioner’s blood, and the negative GC/MS would not invalidate that determination. Spiehler also acknowledges that GC/MS testing produces false negatives. The record contains no basis for concluding any one of these explanations for the discrepancy was less likely than the one identified by the referee.
2. Governing legal principles
The current standard of review for Brady materiality was first articulated in Bagley, supra, 473 U.S. 667, although the United States Supreme Court began developing it in earlier decisions.7 (See Agurs, supra, 427 U.S. at p. 112 [96 S.Ct. at pp. 2401-2402]; Giglio v. United States, supra, 405 U.S. at p. 154 [92 S.Ct. at p. 766].) Recently in Kyles, supra, 514 U.S. 419, the court reemphasized four aspects articulated in Bagley critical to proper analysis of Brady error. First, “[although the constitutional duty is triggered by the potential impact of favorable but undisclosed evidence, a showing of materiality does not require demonstration by a preponderance that disclosure of the suppressed evidence would have resulted ultimately in the defendant’s acquittal (whether based on the presence of reasonable doubt or acceptance of an explanation for the crime that does not inculpate the defendant). [Citations.] Bagley’s touchstone of materiality is a ‘reasonable probability’ of a different result, and the adjective is important. The question is not whether the defendant would more likely than not have received a different verdict with the evidence, but whether in its absence he received a fair trial, understood as a trial resulting in a verdict worthy of confidence.” (Id. at p. 434 [115 S.Ct. at pp. 1565-1566].)
Second, “it is not a sufficiency of evidence test. A defendant need not demonstrate that after discounting the inculpatory evidence in light of the *887undisclosed evidence, there would not have been enough left to convict. The possibility of an acquittal on a criminal charge does not imply an insufficient evidentiary basis to convict. One does not show a Brady violation by demonstrating that some of the inculpatory evidence should have been excluded, but by showing that the favorable evidence could reasonably be taken to put the whole case in such a different light as to undermine confidence in the verdict.” (Kyles, supra, 514 U.S. at pp. 434-435 [115 S.Ct. at p. 1566], fn. omitted.)
Third, “once a reviewing court applying Bagley has found constitutional error there is no need for further harmless-error review.” (Kyles, supra, 514 U.S. at p. 435 [115 S.Ct. at p. 1566].) The one subsumes the other. (Id. at pp. 435-436 [115 S.Ct. at pp. 1566-1567].)
Fourth, while the tendency and force of undisclosed evidence is evaluated item by item, its cumulative effect for purposes of materiality must be considered collectively. (Id. at pp. 436-437 & fn. 10 [115 S.Ct. at p. 1567]; see also Agurs, supra, 427 U.S. at p. 112 [96 S.Ct. at p. 2402], fn. omitted [omission “must be evaluated in the context of the entire record”].)
In Bagley, the court identified another relevant consideration in noting that “an incomplete response to a specific [Brady] request not only deprives the defense of certain evidence, but also has the effect of representing to the defense that the evidence does not exist. In reliance on this misleading representation, the defense might abandon lines of independent investigation, defenses, or trial strategies that it otherwise would have pursued.” (Bagley, supra, 473 U.S. at p. 682 [105 S.Ct. at p. 3384].) Given this possibility, “under the [‘reasonable probability’] formulation the reviewing court may consider directly any adverse effect that the prosecutor’s failure to respond might have had on the preparation or presentation of the defendant’s case. The reviewing court should assess the possibility that such effect might have occurred in light of the totality of the circumstances and with an awareness of the difficulty of reconstructing in a post-trial proceeding the course that the defense and the trial would have taken had the defense not been misled by the prosecutor’s incomplete response.” (Id. at p. 683 [105 S.Ct. at p. 3384]; see, e.g., Payne, supra, 63 F.3d at p. 1209.)
3. Application to the facts
In assessing the materiality of the RIA test result, we begin with the pivotal fact that even without it petitioner offered a diminished capacity defense in an effort to persuade the jury he lacked the ability to premeditate and deliberate at the time of the shooting. Although at trial the claim was based on methamphetamine intoxication, both Beecher and Shinn testified at *888the reference hearing that the positive RIA test would have refocused the defense on PCP intoxication or some combination of the two. The record establishes that the following evidence could have been adduced in support of that defense with reasonable follow-up on the positive RIA.8
Petitioner had an extensive history of drug abuse including PCP. According to Dr. Kaushal Sharma, a forensic psychiatrist who testified for the defense at trial, PCP “causes a wide variety of symptoms including having the effect of producing [a] psychosis-like condition .... [The user] might become . . . extremely violent without provocation.” Sharma also explained that PCP can cause “flashbacks” because when ingested it “becomes attached to the fatty tissue in the brain. . . . And every now and then, sometimes a few months down the line the same PCP which is attached to the fatty tissue in the brain is released into circulation, even if the person has not taken any drugs for months he may again certainly start acting crazy because the PCP which is in the fatty tissue is once again being recirculated in the body through the blood, even though they have not taken any PCP today, they may start. . . acting strange because they have taken PCP two months, two weeks ago.”
In addition to confirming these chemical properties, Pitts noted PCP can cause psychotropic effects at levels below 10 nanograms per milliliter. He also explained that “PCP is a dissociative anesthetic that produces a mental state in which the recipient is oblivious to what is happening in his or her body . . . and can cause a similar dissociative, unconscious mental state that produces random, detached, inappropriate and violent acts. Such reactions to PCP intoxication are not dose related and are variable through time.” According to Pitts, certain aspects of petitioner’s demeanor at the time of his arrest—excessive perspiration and his calmness—as well as his claim of memory loss are consistent with PCP intoxication.9 PCP can also “exacerbate a person’s pre-existing psychotic condition. . . . At a minimum, PCP can impair or prevent a person’s ability to deliberate and premeditate.” Under the governing law at the time of petitioner’s offenses, a defense expert could have rendered such an opinion on the ultimate question of his state of mind. (See, e.g., People v. White (1981) 117 Cal.App.3d 270, 276-277 [172 Cal.Rptr. 612]; cf. Pen. Code, § 29.) In addition, the incidence *889of PCP use was increasing in 1980, and it was used to adulterate other street drugs such as methamphetamine.10
The positive RIA test would have been more than simply the linchpin of this defense. As independent scientific evidence of PCP in petitioner’s blood at the time of the crimes, it would have enhanced the credibility of the other evidence of PCP intoxication it tended to corroborate. (See In re Sixto (1989) 48 Cal.3d 1247, 1264 [259 Cal.Rptr. 491, 774 P.2d 164]; Kyles, supra, 514 U.S. at p. 449, fn. 19 [115 S.Ct. at p. 1573].) We must thus measure not only the likely impact of the RIA itself but also the possible synergistic evidentiary effect it could have generated. (Cf. 514 U.S. at p. 436 [115 S.Ct. at pp. 1566-1567].)
Granted, the prosecution could, and undoubtedly would, have presented rebuttal evidence, including the fact the RIA test was used by the crime lab as a screening mechanism that required confirmation of any positive results. But by definition rebuttal evidence counters—it does not preclude—the defense case-in-chief. Petitioner could also have countered this rebuttal. With respect to the discrepancy between the positive RIA and the negative GC/MS results, all experts agree the two can be reconciled if the amount of PCP in the sample were below the GC/MS cutoff level but sufficient for detection by the RIA, or if the detected substance were a PCP analog for which the GC/MS was not calibrated.11 Spiehler’s testimony did not refute these possibilities. Moreover, she essentially confirmed a 98 percent correlation between RIA and GC/MS results, consistent with Pitts’s statistic. Thus, notwithstanding the screening function of the RIA, the prosecution’s rebuttal would only have resulted in a battle of the experts with significant points of agreement between the two sides.
Respondent does not seriously dispute the conclusion petitioner could have presented a credible diminished capacity defense based on PCP intoxication. In particular, he submits no evidence countering Pitts’s declaration on the ultimate issue of petitioner’s state of mind under the influence of PCP and his lack of ability to premeditate and deliberate. While emphasizing the *890RIA’s function as a screening device, respondent does not claim the test would have been inadmissible for this or any other reason; nor does the record support such a finding.12 Accordingly, the assessment of its evidentiary weight must be left to the trier of fact in light of all other relevant evidence.
As the Supreme Court noted in Kyles, “[t]he inconclusiveness of the physical evidence does not, to be sure, prove Kyles’s innocence, and the jury might have found the eyewitness testimony of [two prosecution witnesses] sufficient to convict, even though less damning to Kyles than that of [two other prosecution witnesses subject to undisclosed impeachment]. But the question is not whether the State would have had a case to go to the jury if it had disclosed the favorable evidence, but whether we can be confident that the jury’s verdict would have been the same.” (Kyles, supra, 514 U.S. at p. 453 [115 S.Ct. at p. 1575], fn. omitted.) When the verdict depends upon the resolution of conflicting evidence, that task is not for a reviewing court. (See ibid.)
In addition, the positive RIA would have diffused or negated some of the rebuttal the prosecution did present. Evidence of petitioner’s demeanor at the time of his arrest was offered because it was inconsistent with the claim of methamphetamine use. As noted, however, Pitts found it consistent with PCP intoxication. More significantly, petitioner would have been able to challenge, with contrary scientific evidence, the testimony of Bonnie Driver that she determined PCP was not present in the blood sample. Nor could the prosecutor have argued strenuously and without contradiction, “So what independent evidence do you have of drug use in this defendant? Independent evidence of drug use this night. You have none. In fact, the only independent evidence of drug use on this night is none. Blood, no drugs in the blood. Conduct? No conduct consistent.” The prosecutor used the GC/MS to make negative reference to Sharma’s testimony “that long after the ingestion of. . . PCP you can determine its presence, [f] The fact which exists in this particular case is that we don’t have those—no evidence of *891those being in the defendant’s blood.” He also questioned Sharma’s opinion as to petitioner’s diminished capacity in light of the negative results. (Cf. Kyles, supra, 514 U.S. at p. 444 [115 S.Ct. at p. 1571].) As in Kyles, “[b]ecause the State withheld evidence, its case was much stronger, and the defense case much weaker, than the full facts would have suggested.” (Id. at p. 429 [115 S.Ct. at p. 1563].)
At this juncture, we must keep in mind petitioner does not have to establish his defense would have ultimately succeeded. (Kyles, supra, 514 U.S. at p. 434 [115 S.Ct. at pp. 1565-1566].) We must also factor other considerations into the materiality equation: Diminished capacity “only needs to be established sufficiently to create a reasonable doubt of the defendant’s guilt when taken into consideration with all the evidence in the case.” (People v. Wells (1938) 10 Cal.2d 610, 622 [76 P.2d 493].) Moreover, petitioner did not seek to avoid guilt entirely but only to negate premeditation and deliberation. At trial, his mental state was the only truly contested issue, which the jury deliberated for a day and a half, even without independent scientific evidence of intoxication.
On this record, we conclude the positive RIA finding was material within the meaning of Brady, supra, 373 U.S. 83. Its nondisclosure prevented petitioner from presenting a credible defense of diminished capacity, thereby denying him a fair trial. As outlined above, this “favorable evidence could reasonably be taken to put the whole case in such a different light as to undermine confidence in the verdict.” (Kyles, supra, 514 U.S. at p. 435 [115 S.Ct. at p. 1566], fn. omitted; see Bagley, supra, 473 U.S. at p. 678 [105 S.Ct. at p. 3381].)
Although we conclude the prosecution failed to disclose exculpatory evidence, we further determine the materiality of that evidence is limited to petitioner’s capacity to premeditate and deliberate; and petitioner does not argue otherwise. He is thus entitled to relief only with respect to the first degree murder conviction, special circumstance finding, and death penalty.
The petition for writ of habeas corpus is granted. The judgment of conviction is vacated, and petitioner is remanded to the Superior Court of Orange County. Upon finality of this decision, the clerk shall remit a certified copy of this opinion to the superior court for filing, and respondent shall serve another copy thereof on the prosecuting attorney in conformity with Penal Code section 1382, subdivision (a)(2). If the People do not elect to bring petitioner to trial within 60 days after service pursuant to Penal *892Code section 1382, subdivision (a)(2), the trial court shall enter judgment reflecting a conviction for second degree murder and shall sentence petitioner accordingly. (See Pen. Code, § 1484; In re Bower (1985) 38 Cal.3d 865, 880 [215 Cal.Rptr. 267, 700 P.2d 1269].)
Mosk, Acting C. J., Werdegar, J., and Crosby, J.,* concurred.