FOR PUBLICATION
  UNITED STATES COURT OF APPEALS
       FOR THE NINTH CIRCUIT

WILLIAM G. OSBORNE,                      
                  Plaintiff-Appellee,
                 v.                             No. 06-35875
DISTRICT ATTORNEY’S                              D.C. No.
                                              CV-03-00118-RRB
OFFICE FOR THE THIRD JUDICIAL
DISTRICT; ADRIENNE BACHMAN,*                     OPINION
District Attorney,
             Defendants-Appellants.
                                         
        Appeal from the United States District Court
                 for the District of Alaska
        Ralph R. Beistline, District Judge, Presiding

                  Argued and Submitted
        October 10, 2007—San Francisco, California

                       Filed April 2, 2008

      Before: Alfred T. Goodwin, Melvin Brunetti, and
            William A. Fletcher, Circuit Judges.

                   Opinion by Judge Brunetti




  *Adrienne Bachman is substituted for former District Attorney Susan
A. Parkes as appellant pursuant to Fed. R. App. P. 43(c)(2).

                               3359
            OSBORNE v. DISTRICT ATTORNEY’S OFFICE       3363


                        COUNSEL

Nancy R. Simel, Assistant Attorney General, Anchorage,
Alaska, for the defendant-appellant.

Peter J. Neufeld and Colin Starger, Innocence Project, Benja-
min N. Cardozo School of Law, New York, New York; Ran-
3364          OSBORNE v. DISTRICT ATTORNEY’S OFFICE
dall S. Cavanaugh, Kalamarides & Lambert, Anchorage,
Alaska; and Robert C. Bundy, Dorsey & Whitney LLP,
Anchorage, Alaska, for the plaintiff-appellee.


                              OPINION

BRUNETTI, Circuit Judge:

  William Osborne, an Alaska prisoner, brought this action
under 42 U.S.C. § 1983 to compel the District Attorney’s
Office in Anchorage to allow him post-conviction access to
biological evidence—semen from a used condom and two
hairs—that was used to convict him in 1994 of kidnapping
and sexual assault. Osborne, who maintains his factual inno-
cence, intends to subject the evidence, at his expense, to STR
and mitochondrial DNA testing, methods that were unavail-
able at the time of his trial and are capable of conclusively
excluding him as the source of the DNA.

   In a prior appeal, Osborne v. District Attorney’s Office, 

, 1056 (9th Cir. 2005) (hereinafter Osborne I),1 we
held that Heck v. Humphrey does not bar Osborne’s § 1983
action because, even if successful, it will not necessarily dem-
onstrate the invalidity of his conviction. We also remanded
for the district court to address in the first instance whether
the denial of access to the evidence violates Osborne’s feder-
ally protected rights.

   In this post-remand appeal, we affirm the judgment of the
district court that, under the unique and specific facts of this
case and assuming the availability of the evidence in question,
Osborne has a limited due process right of access to the evi-
dence for purposes of post-conviction DNA testing, which
  1
    There are three prior “Osborne” appellate opinions, one by this court
and two by the Alaska Court of Appeals. The two state opinions are both
titled Osborne v. State and will be referred to as State I and II.
            OSBORNE v. DISTRICT ATTORNEY’S OFFICE           3365
might either confirm his guilt or provide strong evidence upon
which he may seek post-conviction relief.

                               I

                               A

   Following a March 1994 jury trial in Alaska Superior
Court, Osborne was convicted of kidnapping, assault, and
sexual assault, and was sentenced to 26 years imprisonment,
with 5 years suspended. The charges arose from a March
1993 incident in which the victim, a female prostitute named
K.G., after agreeing to perform fellatio on two male clients,
was driven to a secluded area of Anchorage and brutally
attacked. See generally Jackson v. State, Nos. A-5276, A-
5329, 

, at *1 (Alaska Ct. App. Feb. 7,
1996) (consolidated direct appeal).

   At gunpoint, K.G. was forced to perform fellatio on the
driver while the passenger vaginally penetrated her with his
finger and penis. The driver did not wear a condom, but the
passenger wore a blue condom that K.G. had brought with
her. When K.G. later refused their orders to get out of the car,
the driver hit K.G. in the head with the gun, and at the driver’s
urging the passenger choked her. K.G. eventually attempted
to flee, but her attackers pursued and beat her with an axe
handle. As she lay in the snow in the fetal position and played
dead, she heard the gun fire and felt a bullet graze her head.
Though she could not see her attackers’ faces, judging from
their pants and footwear she believed it was the passenger
who shot her. The attackers then partially covered K.G. with
snow and fled in the car, leaving her for dead.

   K.G. heard the car drive away but continued to lie under
the snow until she was sure her attackers had gone. She then
got up, walked to the main road, flagged down a passing car,
told its occupants what had happened, and—hoping to avoid
the police—asked only for a ride home. The following day,
3366        OSBORNE v. DISTRICT ATTORNEY’S OFFICE
however, a neighbor of one of the car’s occupants notified the
police, who contacted K.G. Though initially uncooperative,
K.G. eventually described the incident.

   K.G. underwent a physical examination, during which hair
and blood samples were collected. A vaginal examination was
not performed, however, because the passenger-rapist had
worn a condom and K.G. had bathed repeatedly since the
attack. At the crime scene, Anchorage Police recovered from
the snow a used blue condom, part of a condom wrapper, a
spent shell casing, and two pairs of K.G.’s grey knit pants
stained with blood. The blue condom and shell casing were
found “very near” each other and in close proximity to bloody
patches of snow and the disturbed berm of snow where K.G.
had been partially buried. A layer of new snow, which had
fallen the morning after the attack, aided the police in distin-
guishing between tire tracks made the night before by the
assailants’ car and tracks made the following day by two
known vehicles. Those cars were owned by K.G.’s rescuers
and their neighbor, who had visited the crime scene the day
after the incident before contacting the police.

   A week later, military police stopped Dexter Jackson for a
traffic infraction. When Jackson opened his glove box to
retrieve his registration, the officer spotted a gun case, which
proved to hold a .380 caliber pistol. A further search of the car
revealed a box of ammunition and a pocketknife. Observing
that Jackson, his car, and his passenger at the time (who was
not Osborne) resembled composite sketches that had been cir-
culated after the assault on K.G., the military police contacted
the Anchorage Police, whom Jackson told that Osborne was
his accomplice on the night of the assault.

  K.G. later identified Jackson and Osborne from photo-
graphic arrays. In identifying Jackson’s accomplice, K.G.
indicated that Osborne’s and another person’s photos were the
“most familiar” to her and Osborne was “most likely” to have
been the passenger who raped and shot her. K.G. also identi-
               OSBORNE v. DISTRICT ATTORNEY’S OFFICE                    3367
fied Jackson’s car, and the police matched tire tracks at the
crime scene to Jackson’s car. K.G. also identified the pocket-
knife found in Jackson’s car as hers, and ballistics tied the
spent shell casing found at the crime scene to Jackson’s pistol.

   The State’s crime lab subjected sperm found in the used
condom to “DQ Alpha” DNA testing, which, similar to ABO
blood typing, reveals the alleles present at a single genetic
locus. The results excluded K.G., Jackson, and James Hunter
(presumably Jackson’s passenger when he was arrested), and
showed that the sperm had the same DQ Alpha type as
Osborne. That same DQ Alpha type is shared, however, by
14.7 to 16 percent of African Americans and thus can be
expected in one of every 6 or 7 black men.

   A DNA testing method called “RFLP,” which was rela-
tively more discriminating than DQ Alpha typing but, accord-
ing to the State, “not quite as discriminating as the testing
[Osborne] now seeks to conduct,” was also available pre-trial
but was not conducted on the sperm.2 The State’s crime lab
expert considered sending out the sample for more discrimi-
nating testing, which was then available through the FBI, but
did not because, at least at that time, more discriminating test-
ing required a better quality sample than was provided in the
condom and the expert “felt that the sample was degraded.”
Defense counsel also considered and rejected the option.
   2
     The State’s concessions that the RFLP DNA testing available pre-trial
is “not quite as discriminating as” the STR and mitochondrial DNA testing
Osborne now seeks, and that these methods were not available pre-trial,
is an apparent reversal of the State’s representation to the state court that
“the DNA testing that Osborne proposes to perform on this evidence
existed at the time of Osborne’s trial, and . . . Osborne’s trial attorney was
aware of this and consciously chose not to seek more specific testing.”
Osborne v. State, 

, 992 (Alaska Ct. App. 2005). According
to the Superior Court’s dismissal order that was on appeal in that case,
Osborne’s state claims involved (1) ineffective assistance at trial based on
counsel’s failure to pursue “RFLP” testing, and (2) a request for post-
conviction “mitochondrial DNA” testing.
3368        OSBORNE v. DISTRICT ATTORNEY’S OFFICE
Counsel met with the DNA expert from the State crime lab,
reviewed DNA research articles, and conferred with a Fair-
banks public defender who was litigating the scientific basis
of DNA testing. But defense counsel’s explanation for not
pursuing pre-trial RFLP testing differs from the State’s
expert’s reasons. According to her post-conviction affidavit,
counsel disbelieved Osborne’s statement that he did not com-
mit the crime, was concerned about a more inculpatory result
nullifying Osborne’s misidentification defense, and concluded
that “Osborne was in a strategically better position without
RFLP DNA testing,” especially given the inherent uncertainty
in the DQ Alpha results.

   The police also recovered two pubic hairs: one from the
used blue condom, and a second from K.G.’s sweatshirt,
which she had lain on top of during the sexual assault. DQ
Alpha typing of these hairs was unsuccessful, likely because
the sample was too small for analysis. Based on microscopic
analysis, however, the State’s expert opined that both hairs
were “dissimilar” to Jackson and Hunter but were “consis-
tent” with having come from Osborne because they “exhibited
the same microscopic features” as Osborne’s pubic hair sam-
ple. Additional hairs having “negroid features” were also
found on K.G.’s clothing but were inconsistent with any of
the suspects investigated by police.

   Osborne and Jackson were tried jointly before a jury.
Osborne presented alibi and mistaken identity defenses, spe-
cifically arguing that there was too little time for him to have
participated in the crime and pointing out flaws in K.G.’s
identification. K.G. was not wearing her glasses on the night
of the attack. She described the passenger who attacked her
as black, between 25 to 30 years old, 6 feet tall, weighing
180-190 pounds, clean shaven, having his hair shaved on the
sides and longer on top, and not wearing any jewelry.
Osborne actually was 20 years old, weighed 155 pounds, and
had a mustache. K.G.’s identification of Osborne was also
cross-racial, Osborne being black and K.G. being white.
            OSBORNE v. DISTRICT ATTORNEY’S OFFICE          3369
Nonetheless, at trial K.G. pointed to Osborne and identified
him as the passenger who attacked her.

   Besides the biological and victim-eyewitness testimony,
there was also circumstantial evidence of Osborne’s culpabil-
ity. Paper tickets from the Space Station arcade, where
Osborne had been before the crime, were found in Jackson’s
car. One group of witnesses saw Osborne get into Jackson’s
car before the time of the crime, and another group saw
Osborne with Jackson after the crime and reported seeing
blood on Osborne’s clothing. Apparently, no trace evidence of
blood on Osborne’s clothing was admitted at trial, however.

   The jury rejected Osborne’s mistaken identity and alibi
defenses and convicted him of kidnapping, first-degree
assault, and two counts of first-degree sexual assault, although
he was acquitted of two counts of attempted first-degree mur-
der and one count of sexual assault. On direct appeal, the
Alaska Court of Appeals rejected Osborne’s insufficient evi-
dence claim and his other challenges and affirmed his convic-
tion. Jackson, 

, at *7-8. Osborne did not
seek direct review in the Alaska Supreme Court.

                               B

   Osborne next filed an application for post-conviction relief
in Alaska Superior Court, arguing first that his trial counsel
was ineffective for failing to pursue RFLP testing, which was
a potentially more precise type of DNA testing and was avail-
able at the time; and second that he has a due process right,
under either the state or federal constitution, to have evidence
retested using DNA testing methods that were not available
until after his trial. In June 2002, the Superior Court denied
his application. Osborne not only appealed that decision to the
Alaska Court of Appeals, he also subsequently filed the
underlying § 1983 action in federal district court.

  While his state appeal and federal action were pending, in
April 2004 Osborne also applied for discretionary parole with
3370        OSBORNE v. DISTRICT ATTORNEY’S OFFICE
the Alaska Board of Parole. In his written application,
Osborne confessed to participating in the attack on K.G. and
described his actions in detail. He also confessed at his
August 2005 hearing before the Parole Board and stated that
he had told his mother and lawyer the truth about the incident.
But despite his efforts at accepting responsibility, Osborne
was denied parole.

   Osborne also lost his state appeal. In reviewing the Supe-
rior Court’s dismissal of Osborne’s petition for post-
conviction relief, the Alaska Court of Appeals affirmed in
part, remanded in part for further proceedings, and retained
jurisdiction in the interim. Osborne v. State, 

,
995-96 (Alaska Ct. App. 2005) (hereinafter State I). The court
rejected Osborne’s ineffective assistance claim, holding that
he failed to establish deficient performance. Like the Superior
Court, the Court of Appeals noted that trial counsel “re-
searched and considered” RFLP DNA testing, but decided
against it for fear that a more discriminating test would further
inculpate Osborne. That decision, the court concluded, was a
“tactical” one that fell within the permissible range of attor-
ney competence. 

The court made no mention of
the State’s DNA expert’s trial testimony that the sperm sam-
ple was too degraded to even permit RFLP testing.

   As to Osborne’s due process claim, the Court of Appeals
initially observed that a prisoner “apparently” has no federal
due process right to present new post-conviction evidence to
establish a freestanding claim of factual innocence, absent an
underlying constitutional defect at trial. 

995. “[A]s
a matter of Alaska constitutional law,” however, the court was
“reluctant to hold that Alaska law offers no remedy to defen-
dants who could prove their factual innocence,” and it
adopted a three-part test, which had been adopted by several
other states, for cases in which defendants seek post-
conviction DNA testing. 

The court therefore
remanded to the Superior Court for the limited purpose of
determining whether Osborne could satisfy the test and, if so,
            OSBORNE v. DISTRICT ATTORNEY’S OFFICE           3371
whether Osborne’s claim was otherwise procedurally barred
under Alaska law. 

proceedings in Alaska Superior Court were
still pending when we decided Osborne I in September 2005.
Eight months later, the Alaska Superior Court held that
Osborne failed to satisfy the three factors set forth by the
Alaska Court of Appeals in State I and therefore denied
Osborne’s request for DNA testing. The Superior Count
found that (1) “Osborne’s conviction did not rest primarily
upon eyewitness identification evidence,” (2) “no . . . demon-
strable doubt as to Mr. Osborne’s identification [as the perpe-
trator] has been established,” and (3) “[e]ven assuming that
the DNA of either the pubic hair or the semen was tested and
found not to be Mr. Osborne’s, . . . such evidence, if obtained,
would not be conclusively exculpatory.”

   In Osborne v. State, 

, 979-81 (Alaska Ct. App.
2007) (hereinafter State II), the Alaska Court of Appeals
affirmed those three findings and the Superior Court’s under-
lying factual findings and reasoning. Although Osborne sub-
sequently filed a petition for review in the Alaska Supreme
Court, which was pending when we heard oral argument in
this case, that petition has since been denied. Osborne v.
State, No. S-12799 (Alaska Jan. 22, 2008).

                               C

   In June 2003—one year after the Alaska Superior Court
first denied Osborne’s state petition for post-conviction relief,
and one year before Osborne filed his application for discre-
tionary parole in which he provided his written confession—
Osborne filed the underlying action under 42 U.S.C. § 1983
alleging that the Anchorage District Attorney’s Office, then-
District Attorney Susan Parkes, the Anchorage Police Depart-
ment, and then-Chief of Police Walt Monegan violated his
rights under the United States Constitution by denying him
post-conviction access to evidence for DNA testing. Specifi-
3372        OSBORNE v. DISTRICT ATTORNEY’S OFFICE
cally, he alleges violations of: (1) his due process right to
access exculpatory evidence; (2) his due process right to dem-
onstrate actual innocence; (3) his Eighth Amendment right to
be free from cruel and unusual punishment; (4) his right to a
fair clemency hearing; (5) his Sixth Amendment rights to con-
frontation and compulsory process; and (6) his due process
and equal protection rights to meaningful access to the courts.
He requests as a remedy “the release of the biological
evidence”—the semen and pubic hair from the blue condom
and the pubic hair from K.G.’s sweater—and “the transfer of
such evidence for DNA testing.”

   Osborne’s complaint explains that he intends to subject the
evidence to two forms of DNA testing: Short Tandem Repeat
(“STR”) analysis and Mitochondrial DNA (“mtDNA”) analy-
sis. Unlike the DQ Alpha analysis used at trial, which looks
to only one genetic locus, STR analysis examines the alleles
at 13 genetic loci. Thus, it has the power to produce a far
more specific genetic profile—one shared by one in a billion
people, rather than one in 6 or 7. See generally United States
v. Kincade, 

, 818-19 (9th Cir. 2004) (en banc)
(discussing the science and accuracy of STR testing).

   Moreover, if the DNA samples are unsuitable for STR anal-
ysis, Osborne intends to submit them to mtDNA analysis.
STR analysis, like DQ Alpha analysis, examines DNA found
in the nucleus, and is incapable of reaching a result from a
hair sample unless the root or follicle is attached. By contrast,
mtDNA analysis looks to DNA found in the mitochondria,
and does not require the presence of a root or follicle.
Osborne asserts that neither STR nor mtDNA analysis was
available at trial and that he will have the testing performed
at his expense.

   In 2003, the State filed motions to dismiss Osborne’s
§ 1983 action, arguing that Osborne’s simultaneously pending
state petition for post-conviction relief required federal court
abstention under Younger v. Harris, 

(1971), and,
            OSBORNE v. DISTRICT ATTORNEY’S OFFICE          3373
alternatively, that under Heck v. Humphrey, 

(1994), Osborne’s claims are not cognizable in a § 1983
action and may be brought only in a habeas proceeding. The
district court granted the motion, holding that Younger absten-
tion was inapplicable but that dismissal was required under
Heck because Osborne sought to “set the stage” for an attack
on his conviction.

   Osborne appealed, and in Osborne I we reversed, holding
that Heck does not bar a prisoner’s § 1983 action seeking
post-conviction access to biological evidence for purposes of
DNA testing because, even if successful, the action will not
necessarily demonstrate the invalidity of his conviction.
Osborne v. District Attorney’s Office, 

, 1056
(9th Cir. 2005). We observed that “success would yield only
access to the evidence—nothing more,” “there is a significant
chance that the results will either confirm or have no effect on
the validity of Osborne’s confinement” because “further DNA
analysis may prove exculpatory, inculpatory, or inconclu-
sive,” and “even if the results exonerate Osborne, a separate
action—alleging a separate constitutional violation altogether
—would be required to overturn his conviction.” 

55. Expressing no opinion on the merits of Osborne’s claims,
we remanded to the district court to address in the first
instance whether Osborne has been deprived of a federally
protected right. 

   On remand, Osborne moved for summary judgment on his
§ 1983 claims. Days later, Osborne along with the Anchorage
Police Department and Chief of Police Monegan filed a stipu-
lation dismissing those two defendants in exchange for their
withdrawal of opposition to Osborne’s claim for DNA testing
and their agreement to make the evidence available as
required by any final court order or upon direction from the
State of Alaska, Department of Law.

  The remaining “State” defendants—the District Attorney
and the District Attorney’s Office—filed both an opposition
3374        OSBORNE v. DISTRICT ATTORNEY’S OFFICE
to Osborne’s summary judgment motion and a motion to stay
proceedings based on the Colorado River and Pullman
abstention doctrines. See Colo. River Water Conservation
Dist. v. United States, 

(1976); R.R. Comm’n of
Tex. v. Pullman Co., 

(1941). The district court
denied the abstention motion, and the State has not appealed
that denial.

   Months later, and just days after the Alaska Superior Court
issued its Findings on Remand from State I, the State filed a
cross-motion for summary judgment. The State asserted, in
addition to its arguments opposing Osborne’s motion, that the
Superior Court’s factual findings were entitled to preclusive
effect in federal court and otherwise support the State’s posi-
tion.

   This time the district court ruled in Osborne’s favor.
Osborne v. Dist. Attorney’s Office, 

(D.
Alaska 2006). After initially concluding that the Alaska Court
of Appeals’ decision in State I “is not binding upon this
Court,” the district court held that “there does exist, under the
unique and specific facts presented, a very limited constitu-
tional right to the testing sought.” 

(citing Har-
vey v. Horan (Harvey II), 

, 325 (4th Cir. 2002)
(Luttig, J., respecting the denial of rehearing en banc);
Thomas v. Goldsmith, 

, 749-50 (9th Cir. 1992);
and Moore v. Lockyer, No. 04-1952, 

, at
*8, *9 (N.D. Cal. Sept. 23, 2005)). The court continued:

    Significant to this conclusion is the fact that the test-
    ing sought was not available to Plaintiff . . . at the
    time of trial, the fact that the testing sought can be
    easily performed without cost or prejudice to the
    Government, and the fact that the test results can
    either confirm Plaintiff’s guilt or provide evidence
    upon which Plaintiff might seek a new trial. In this
    regard, and although the Court makes no findings as
    to whether Plaintiff would be entitled to a new trial,
            OSBORNE v. DISTRICT ATTORNEY’S OFFICE             3375
    the Court concludes that a favorable result for Plain-
    tiff might have a significant impact on a fact-
    finder[’]s evaluation of guilt or innocence.

       The Court also concludes that equity and funda-
    mental notions of fairness argue in favor of the relief
    Plaintiff seeks; especially, when considered in the
    appropriate context, e.g., the Government has no
    legitimate interest in punishing the innocent.

       The Court’s conclusion in this matter assumes the
    availability of the evidence sought. If the evidence
    were no longer available, for any legitimate reason,
    there exists no right to test it and no basis upon
    which Plaintiff could pursue the issue further.

(footnote omitted).

   We have jurisdiction under 28 U.S.C. § 1291. We review
de novo the district court’s rulings on cross-motions for sum-
mary judgment. Phillips v. Hust, 

, 1075 (9th
Cir. 2007). We also review de novo the legal question of
whether there exists a post-conviction right of access to evi-
dence for DNA testing. Cf. 

Whether the State’s
refusal to grant access to evidence violates Osborne’s consti-
tutional due process right is also reviewed de novo as a mixed
question of fact and law. Cf. United States v. Duff, 

, 177 (9th Cir. 1987).

                               II

   [1] “It is well settled that the government has the obligation
to turn over evidence in its possession that is both favorable
to the accused and material to guilt or punishment.” Pennsyl-
vania v. Ritchie, 

, 57 (1987). Yet Ritchie, like the
rest of the Supreme Court’s cases involving Brady rights,
involved only the right to pre-trial disclosure. See 

States v. Agurs, 

(1976), and Brady v.
3376        OSBORNE v. DISTRICT ATTORNEY’S OFFICE
Maryland, 

, 87 (1963), both of which involved
only pre-trial suppression); cf. 

(describing the duty
to disclose as “ongoing,” yet simultaneously referencing only
reconsideration of disclosure “as the proceedings progress” to
ensure “the fairness of the trial”). The more novel question
presented in this case is whether, and the extent to which, the
Due Process Clause of the Fourteenth Amendment extends
the government’s duty to disclose (or the defendant’s right of
access) to post-conviction proceedings.

   [2] As both parties and the district court have recognized,
this circuit’s closest precedent is Thomas v. Goldsmith, 

, 749-50 (9th Cir. 1992), in which we ordered the
disclosure of potentially-exculpatory semen evidence in a
habeas corpus proceeding where testing of the evidence was
potentially material to a so-called “gateway” showing of
actual innocence. Rather than relying on general discovery
principles or a pre-trial Brady right, we expressly applied
Brady as a post-conviction right, stating:

    [W]e believe the state is under an obligation to come
    forward with any exculpatory semen evidence in its
    possession. See Brady v. Maryland, 

, 87,
    . . . (1963). We do not refer to the state’s past duty
    to turn over exculpatory evidence at trial, but to its
    present duty to turn over exculpatory evidence rele-
    vant to the instant habeas corpus proceeding.

. Although the State contends that
Thomas is distinguishable on a variety of grounds, we find
none of those arguments persuasive and conclude that Thomas
is controlling here despite the factual and procedural differ-
ences.

   [3] For instance, the State argues that Thomas involved pre-
trial suppression in addition to post-conviction suppression,
whereas Osborne had full pre-trial access to the biological
evidence in question for purposes of DNA testing “using the
            OSBORNE v. DISTRICT ATTORNEY’S OFFICE           3377
best technology available at that time.” Moreover, the State
argues, in Thomas the petitioner claimed ineffective assis-
tance based on trial counsel’s failure to test the evidence,
whereas the Alaska state courts have already rejected
Osborne’s similar ineffective assistance claim. But this is all
rather beside the point given our reasoning in Thomas. It is
patent in the above-quoted passage that in granting the peti-
tioner a post-conviction right of access we expressly applied
Brady as a post-trial right and specifically disclaimed reliance
on a pre-trial Brady violation or any other pre-trial violation.
Pursuant to Thomas, the more material facts here are that the
STR and mtDNA testing methods now being requested were
in fact not accessible to Osborne pre-trial, whether due to sup-
pression, ineffective assistance of counsel, or historical scien-
tific limitations; such methods are far more discriminating
than the testing methods that were available pre-trial, such
that only now can Osborne be identified or excluded as the
source of the DNA to a virtual certainty; and, for the same
reasons, these newly available methods have the potential to
provide strong evidence upon which Osborne might seek
post-conviction relief. Cf. Riofta v. State, 

, 200,
203 (Wash. Ct. App. 2006) (distinguishing Thomas because
Riofta had pre-trial access to the evidence and to testing “of
equal accuracy” to that sought post-trial; also distinguishing
a New Jersey case in which “DNA testing was not widely
accepted and was expensive at the time of the . . . defendant’s
trial”).

   At the time of briefing, the State further contended that the
“Catch-22” that was present in Thomas does not exist here
because Osborne had an alternative avenue for relief in state
court, which he was at that time pursuing. As the Alaska
Supreme Court has since denied Osborne’s petition for review
in Osborne v. State, No. S-12799 (Alaska Jan. 22, 2008), the
State’s argument is probably moot. But even if Osborne still
had some available state remedy that he could pursue, the
Catch-22 would remain. In both state and federal court the
State has opposed Osborne’s access-to-evidence claim based
3378         OSBORNE v. DISTRICT ATTORNEY’S OFFICE
on the argument that Osborne cannot prove his actual inno-
cence; yet Osborne needs access to the evidence to make that
very showing. Cf. 

. Although the
State further contends that even assuming favorable test
results Osborne could not make a sufficient showing of actual
innocence, that argument is out of place. The argument is cer-
tainly relevant to the ultimate question of whether, given the
state of the evidentiary record, the requested DNA testing
would be sufficiently material to require disclosure in this
case. See infra Part IV. But the State’s argument is irrelevant
to the instant threshold issue of whether Osborne can even
begin to invoke Thomas as establishing a post-conviction
Brady right. Moreover, to the extent the existence of parallel
state litigation might have raised abstention concerns, the
State’s motions to dismiss or stay proceedings based on Youn-
ger, Colorado River, and Pullman abstention were all denied
by the district court, and the State failed to reassert its absten-
tion arguments in either Osborne I or the instant appeal.

   [4] The State finally contends that Thomas is distinguish-
able because Osborne does not have an ongoing federal
habeas action to which the requested DNA testing would be
material. According to the State, such a limitation on Thomas
is necessary to prevent Brady from applying “in the abstract
and in perpetuity.” We disagree and reject the notion that
Thomas’ recognition of a post-conviction Brady right is inap-
plicable per se in the absence of an ongoing habeas proceed-
ing.

   While recognizing that Osborne’s sole purpose in bringing
the underlying § 1983 action is to obtain post-conviction
access to potentially exculpatory evidence and thereby “to
‘facilitate’ or ‘set the stage’ for a future attack on his convic-
tion,” in Osborne I we specifically rejected the State’s posi-
tion that a post-conviction access-to-evidence claim must be
brought in habeas rather than under § 

.
Yet that would be the necessary consequence of the State’s
proposed limitation of Thomas. The purportedly mandatory
            OSBORNE v. DISTRICT ATTORNEY’S OFFICE         3379
habeas action would render any § 1983 action for access to
evidence superfluous, leaving the habeas action as the exclu-
sive vehicle for asserting Thomas rights. Thus, the State’s
proposed limitation of Thomas does little more than rehash in
different clothing the already rejected contentions that
Osborne has failed to state a claim under § 1983 and “that a
claim which can be brought in habeas must be brought in
habeas.” 

access-to-evidence claim has been
pursued without undue delay and is specifically intended to
support an application for post-conviction relief. Osborne has
declared in his § 1983 complaint his intention, once he
obtains the requested DNA testing, to file a petition for post-
conviction relief based on a “freestanding” claim of actual
innocence—i.e., despite the lack of any constitutional error at
his trial, his incarceration is unconstitutional due to actual
innocence—which Osborne might bring in either state or fed-
eral court.

   Under Alaska law Osborne appears to have at least a poten-
tially viable opportunity of bringing a freestanding actual
innocence claim in a second or successive petition for post-
conviction relief. However unfavorable the court’s rejection
of Osborne’s first petition in State II may be, his state law
options are not foreclosed entirely. In a concurrence, a two-
judge majority stated that, regardless of any otherwise appli-
cable procedural bars (presumably including those against
successive petitions), the Alaska Constitution might require
court intervention if a defendant were to present “clear
genetic evidence of [his] innocence,” and “[i]f Osborne could
show that he were in fact innocent, it would be unconsciona-
ble to punish him.” State 

(Mannheimer,
J., joined by Coats, C.J., concurring). Thus, despite the
improbability of success on the merits given the court’s find-
ings regarding the expected immateriality of the requested
DNA testing, if Osborne’s hunt for new evidence exceeds the
state courts’ expectations (which remains a possibility, see
3380        OSBORNE v. DISTRICT ATTORNEY’S OFFICE
infra Part IV.B) his actual innocence claim might be cogniza-
ble under the authority of Judge Mannheimer’s concurrence.

   Federal law presents a similar opportunity. The State would
take the position that a freestanding actual innocence claim is
not cognizable under federal law; however, the State also con-
cedes that it is presently an open question. In Herrera v. Col-
lins, 

, 417 (1993), the Supreme Court assumed
without deciding that such a claim is possible. And in House
v. Bell, 

(2006), the Court again declined to
resolve “the question left open in Herrera” in such a way as
to suggest that it will remain unanswered until it is squarely
presented by a petitioner actually making a persuasive show-
ing of actual innocence. 

(concluding that “what-
ever burden a hypothetical freestanding innocence claim
would require, this petitioner has not satisfied it”). The same
might be said of the question of whether there should be a dis-
tinction between capital and non-capital cases, although Her-
rera did suggest equal treatment. See 

(“[W]e have ‘refused to hold that the fact that a death
sentence has been imposed requires a different standard of
review on federal habeas corpus.’ ” (quoting Murray v. Giar-
ratano, 

, 9 (1989) (plurality opinion))).

  In this circuit we not only have assumed that freestanding
innocence claims are possible but also have articulated a
minimum standard: “a habeas petitioner asserting a freestand-
ing innocence claim must go beyond demonstrating doubt
about his guilt, and must affirmatively prove that he is proba-
bly innocent.” Carriger v. Stewart, 

, 476 (9th
Cir. 1997) (en banc); see also Jackson v. Calderon, 

, 1164-65 (9th Cir. 2000).

   In resolving the instant appeal, we need not decide the open
questions surrounding freestanding actual innocence claims.
Instead, we assume for the sake of argument that such claims
are cognizable in federal habeas proceedings in both capital
and non-capital cases under the standard set forth in Carriger.
             OSBORNE v. DISTRICT ATTORNEY’S OFFICE           3381
Such a claim is not yet before us; it would require the filing
of “a separate action—alleging a separate constitutional viola-
tion altogether.” Osborne 

. Also, even
where an actual innocence claim has been filed, Herrera,
House, Carriger, and Jackson all support the practice of first
resolving whether a petitioner has made an adequate evidenti-
ary showing of actual innocence before reaching the constitu-
tional question of whether freestanding innocence claims are
cognizable in habeas. And under Majoy v. Roe, 

(9th Cir. 2002), the testing of potentially exculpatory evidence
may be given precedence over the consideration of even juris-
dictional questions involving pure issues of law.

   [5] In Majoy, we held that the district court should first
order that testing be conducted, hold an evidentiary hearing to
permit full development of the facts supporting a “gateway”
actual innocence claim under Schlup v. Delo, 

(1995), and determine whether that claim was factually meri-
torious, all before considering whether the claim was jurisdic-
tionally barred. 

. We reasoned that
the issue of whether a gateway actual innocence claim pro-
vides a constitutional exception to AEDPA’s statute of limita-
tions was such an “important legal question” that it “is not
appropriately addressed by us in a hypothetical context,” par-
ticularly given the “ ‘fact-intensive nature of this [actual inno-
cence] inquiry, together with the District Court’s ability to’ ”
obtain the necessary evidence. 

(quoting 

). By extension, until Osborne has actually brought
an actual innocence claim and has been given the opportunity
to develop the facts supporting it, Osborne’s access-to-
evidence claim may proceed on the well-established assump-
tion that his intended freestanding innocence claim will be
cognizable in federal court.

   There are two notable differences between the out-of-
circuit and district court cases that do not recognize a post-
conviction right of access to evidence and those that do. First,
the former tend to conflate the right of access to evidence
3382        OSBORNE v. DISTRICT ATTORNEY’S OFFICE
with the ultimate right to habeas relief. See, e.g., State 

, 993 (noting that Osborne claims a right of access
to evidence, but analyzing whether Herrera permits a claim
of actual innocence based on new evidence); Harvey v. Horan
(Harvey I), 

, 375 (4th Cir. 2002) (dismissing
under Heck). The latter distinguish the two actions. See, e.g.,
Harvey v. Horan (Harvey II), 

, 322-24 (4th Cir.
2002) (Luttig, J., respecting the denial of rehearing en banc).

   Second, courts denying a post-conviction right of access
read Brady and its progeny as applying only as a trial right.
E.g., Grayson v. King, 

, 1337, 1342 (11th Cir.
2006) (ultimately limiting its holding to “the particular cir-
cumstances of this case”); Harvey 

. By
contrast, courts recognizing a post-conviction right have done
so not necessarily based on Brady itself but based on the due
process principles that motivated Brady, including fundamen-
tal fairness, the prosecutor’s obligation to do justice rather
than simply obtain convictions, and the “constitutional imper-
atives of ‘protecting the innocent from erroneous conviction
and ensuring the integrity of our criminal justice system.’ ”
Wade v. Brady, 

, 246 (D. Mass. 2006)
(quoting California v. Trombetta, 

, 485 (1984));
see also Harvey 

(Luttig, J., respecting
the denial of rehearing en banc); Moore v. Lockyer, 

, at *8 (N.D. Cal. Sept. 23, 2005), appeal docketed,
No. 06-15016 (9th Cir. argued Oct. 15, 2007).

   [6] This circuit has already staked out positions on both of
these issues. In Osborne 

, we drew a
sharp distinction between access-to-evidence and habeas
claims in holding that Osborne’s access claim is not barred by
Heck v. Humphrey. And in 

, we
expressly applied Brady as a post-conviction right in a habeas
proceeding based on the requirements of fundamental fair-
ness. Faced now with the argument that Thomas should be
limited to cases with ongoing habeas petitions, we reject that
view and hold that Osborne is entitled to assert in this § 1983
             OSBORNE v. DISTRICT ATTORNEY’S OFFICE           3383
action the due process right to post-conviction access to
potentially exculpatory DNA evidence that we recognized in
Thomas.

                               III

   The State alternatively contends that, even if there is a post-
conviction right of access, the scope of the right recognized
by the district court is too broad. It proposes a narrower stan-
dard for judging the materiality of evidence favorable to
Osborne. In the State’s view, before it is obligated to disclose
any evidence post-conviction, Osborne should be required to
satisfy the extraordinarily high standard of proof that applies
to freestanding claims of actual innocence. Thus, under our
case law, Osborne would be required to “go beyond demon-
strating doubt about his guilt, and must affirmatively prove
that he is probably innocent.” 

; see
also 

. We disagree.

   The fundamental flaw in the State’s position is that it effec-
tively equates Osborne’s access-to-evidence claim with a
habeas claim. Under the State’s conception, Osborne would
have to satisfy the same actual innocence standard in order to
merely obtain potentially exculpatory evidence as he would in
order to actually overturn his conviction. The only difference
is in the factual predicate: we would analyze Osborne’s access
claim taking into account the eventuality that further DNA
testing will be favorable to Osborne, whereas habeas relief
would require the actual results. The application of Carriger’s
legal standard, however, would otherwise be identical.

  [7] The State is getting ahead of itself. Obtaining post-
conviction access to evidence is not habeas relief. And requir-
ing Osborne to demonstrate that he would be entitled to
habeas relief if the test results are favorable in order even to
conduct such testing is fundamentally inconsistent with
Brady, Thomas, and Osborne I. Under these authorities, the
most stringent materiality standard for simply obtaining post-
3384        OSBORNE v. DISTRICT ATTORNEY’S OFFICE
conviction access to evidence must be more lenient than the
standard of proof Osborne will ultimately have to satisfy in
order to obtain habeas relief.

   While Brady ensures a fair trial, a defendant’s right to pre-
trial disclosure under Brady is not conditioned on his ability
to demonstrate that he would or even probably would prevail
at trial if the evidence were disclosed. Rather, disclosure is
required if there is a “reasonable probability” of a more favor-
able result at trial, and “the adjective is important.” Kyles v.
Whitley, 

, 434 (1995). As the Supreme Court
explained in Kyles, “a showing of materiality does not require
demonstration by a preponderance that disclosure of the sup-
pressed evidence would have resulted ultimately in the defen-
dant’s acquittal.” 

it “a sufficiency of evidence test.
A defendant need not demonstrate that after discounting the
inculpatory evidence in light of the undisclosed evidence,
there would not have been enough left to convict.” 

35. Instead, materiality for Brady purposes is established “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.” 

   In Thomas, we extended Brady as supporting a habeas peti-
tioner’s post-conviction right to obtain access to semen evi-
dence for purposes of DNA 

.
While we did not expressly adopt Brady’s materiality stan-
dard for post-conviction access claims or otherwise define the
applicable materiality standard, our decision was in general
accordance with the principles underlying Brady materiality.
Despite the fact that the petitioner ultimately needed the
semen evidence in order to establish a Schlup “gateway”
claim of actual innocence for purposes of overcoming a pro-
cedural default, we did not condition the petitioner’s right of
access on his ability to first demonstrate that favorable test
results would in fact enable the petitioner to make a colorable
showing of actual innocence. Instead, we merely noted that
“[a] semen sample, or tests thereof, might enable him to make
            OSBORNE v. DISTRICT ATTORNEY’S OFFICE           3385
such a showing,” “[i]n light of the obvious exculpatory poten-
tial of semen evidence in a sexual assault case.” 

750 n.2 (emphasis added). We put off analyzing the complete
factual basis for petitioner’s actual innocence claim until the
evidence, if it even still existed, was actually produced and
tested. 

   Similarly, in Jones v. Wood, 

(9th Cir. 1997),
we held that a habeas petitioner was statutorily entitled to
post-conviction testing of physical evidence and other discov-
ery because such “discovery is essential for Jones to develop
fully his ineffective assistance of counsel claim,” and “the test
results may establish the prejudice required to make out such
a claim.” 

(emphasis added). And in Majoy, we
found sufficient “the distinct possibility that given the oppor-
tunity, Majoy may be able to muster a plausible factual case
meeting the exacting gateway standard [of actual innocence]
established by the Supreme Court in 

(emphasis added).

   In Osborne I, we did not address the standards Osborne
might have to satisfy in order to prevail on his access-to-
evidence claim. But our decision that Heck v. Humphrey does
not bar Osborne’s § 1983 action was based on the rationale
that Osborne’s access-to-evidence claim is not the functional
equivalent of a habeas petition because it would not necessar-
ily demonstrate the invalidity of his conviction. Indeed, “a
separate action—alleging a separate constitutional violation
altogether—would be required to overturn his conviction.”
Osborne 

. Nonetheless, rather than allow
the objective facts to come to light so that Osborne can actu-
ally file his actual innocence claim and support it with hard
evidence, the State effectively would have us fully analyze
that same claim as a hypothetical by adopting the same habeas
standard in the instant action. In accordance with the cases
discussed above, we decline to do so.

  For his part, Osborne contends that the ordinary pre-trial
Brady materiality standard is the most appropriate standard
3386        OSBORNE v. DISTRICT ATTORNEY’S OFFICE
for evaluating a prisoner’s post-conviction right of access to
evidence. Thus, he proposes that we require only a reasonable
probability that, had the evidence been disclosed to the
defense, the result of his trial would have been different. See

; United States v. Bagley, 

, 682 (1985).

   As it turns out, however, Osborne’s case for disclosure is
so strong on the facts that his proposed legal standard, which
would give him the benefit of both the presumption of inno-
cence and the requirement of jury unanimity for conviction,
sets the bar far lower than what he is able to show in this case.
Wherever the bar is, he crosses it. This case therefore does not
require us to determine the full breadth of post-conviction
Brady rights. The precise height at which the materiality bar
should be set is largely an academic question, which we may
leave to another day and another case that truly presses the
issue. Resolving the instant case requires us to determine only
that Osborne’s showing of materiality is sufficient to require
disclosure, and not whether the same due process right might
be invoked upon a lesser showing.

   [8] In accordance with that reservation and our analysis in
rejecting the State’s proposal as setting the bar too high, we
hold that the standard of materiality applicable to Osborne’s
claim for post-conviction access to evidence is no higher than
a reasonable probability that, if exculpatory DNA evidence
were disclosed to Osborne, he could prevail in an action for
post-conviction relief. Taking into account Osborne’s
declared intention to file a freestanding claim of actual inno-
cence, materiality would be established by a reasonable prob-
ability that Osborne could “affirmatively prove that he is
probably innocent.” 

. And to para-
phrase the Supreme Court’s definition of “reasonable proba-
bility,” this materiality standard does not require a
demonstration by a preponderance that disclosure of the DNA
evidence will ultimately enable Osborne to prove his inno-
cence. See 

. The question is not whether
             OSBORNE v. DISTRICT ATTORNEY’S OFFICE           3387
Osborne would more likely than not be granted habeas relief
with the evidence, but whether in the absence of the DNA evi-
dence Osborne would receive a fair habeas hearing, under-
stood as a hearing resulting in a judgment “worthy of
confidence.” 

next, considered in light of the particular facts
of this case, this standard is satisfied by the potential proba-
tive value of the DNA evidence to which Osborne seeks
access.

                               IV

                                A

   As an initial matter, the State contends that in assessing the
potential materiality of further DNA testing, this court should
give preclusive effect to three factual findings made by the
Alaska Superior Court, and later affirmed by the Alaska Court
of Appeals in State II, when the court rejected Osborne’s
request for DNA testing under state law. Those findings are
that (1) Osborne’s conviction did not rest primarily on eyewit-
ness identification evidence, (2) there was no demonstrable
doubt concerning the accuracy of the victim’s identification of
Osborne as the perpetrator, and (3) additional DNA testing
would not conclusively establish Osborne’s innocence. State

.

   [9] Whether the state court’s factual findings have preclu-
sive effect in this federal proceeding is a mixed question of
law and fact that we review de novo. See Littlejohn v. United
States, 

, 919 (9th Cir. 2003). In § 1983 actions,
we apply state law to determine the preclusive effect of a state
court’s findings. Heck v. Humphrey, 

, 480 n.2
(1994). Under Alaska law, factual findings are entitled to pre-
clusive effect if the party against whom estoppel is asserted
was a party to or in privity with a party to the prior action, the
issue to be precluded is identical to that decided in the first
3388        OSBORNE v. DISTRICT ATTORNEY’S OFFICE
action, and the issue in the first action was resolved by a final
judgment on the merits. Holmberg v. State, 

, 827
(Alaska 1990). The parties here raise no dispute regarding the
privity and final judgment requirements. They dispute only
whether the issues decided by the Alaska courts are identical
to the issues raised by Osborne’s access-to-evidence claim
under federal law.

   The state court’s first and second findings regarding the
evidence supporting Osborne’s identification as the perpetra-
tor are certainly relevant to our inquiry, but only insofar as
such evidence is part of the broader evidentiary backdrop
against which the materiality of exculpatory DNA tests is to
be analyzed. The state court’s findings fundamentally differ
from our materiality inquiry, however, in that they are exclu-
sively historical, focusing only on the state of the evidence as
it existed at trial and whether that trial record would lead one
to question the integrity of that evidence, much like a suffi-
ciency of the evidence inquiry under Jackson v. Virginia, 

, 324 (1979). By contrast, the materiality of sup-
pressed evidence and the viability of an actual innocence
claim based on new evidence are far more comprehensive and
forward-looking inquiries that do not “turn on discrete find-
ings regarding disputed points of fact.” 

. “One does not show a Brady violation by demonstrat-
ing that some of the inculpatory evidence should have been
excluded [or was inherently flawed], 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
a decision made without the evidence. 

.
In assessing whether DNA results favorable to Osborne would
be material to an actual innocence claim, we “must consider
all the evidence, old and new, incriminating and exculpatory,”
and “make a probabilistic determination about what [a reason-
able fact-finder] would do.” 

(internal
quotation marks omitted).
               OSBORNE v. DISTRICT ATTORNEY’S OFFICE                   3389
   [10] Thus, only the state court’s third finding—which is
essentially its “materiality” finding under Alaska law—is in
play here. But that finding is also not entitled to preclusive
effect in this case because it was made in conformity with a
materiality standard under state law that is more stringent than
any standard this court would apply under federal law.
Whereas the Alaska courts inquired as to whether DNA tests
excluding Osborne as the source of the genetic material would
“conclusively establish Osborne’s innocence,” State 

(emphasis added),3 we have determined that mate-
riality under federal law requires Osborne to demonstrate, at
most, only a reasonable probability that with favorable DNA
test results he could affirmatively prove that he is probably
innocent. Cf. 

, 553-54 (holding that,
although there was not “conclusive exoneration,” the peti-
tioner satisfied the Schlup standard for a gateway claim of
actual innocence—“that more likely than not, in light of the
new evidence, . . . any reasonable juror would have reason-
able doubt”). Indeed, Alaska’s materiality standard for merely
obtaining post-conviction access to evidence is more stringent
than even this circuit’s standard for obtaining habeas relief
  3
   The Alaska Court of Appeals originally phrased the broader issue
under Alaska law as “whether further DNA testing . . . would likely be
conclusive on the issue of Osborne’s guilt or innocence.” State 

(emphasis added). But that inquiry undergoes a significant
metamorphosis as the opinion progresses. After the court acknowledges
that further DNA testing could produce three different results—
inculpatory, inconclusive, or exculpatory—from that point forward the
court presumes test results favorable to Osborne and the word “likely” dis-
appears from its statement of the legal standard. 

inquiry ulti-
mately becomes: “assuming that this third alternative came to pass—i.e.,
assuming that a more discriminating DNA test showed that the genetic
material did not come from Osborne—would this test result be conclusive
evidence of Osborne’s innocence?” 

if we have misread
or were to ignore this shift, we would reach the same result. First, it is
unclear whether “likely” equates to “reasonable probability” within the
meaning of Bagley and Kyles. And second, a freestanding actual inno-
cence claim under Carriger requires only an affirmative showing of prob-
able innocence, not conclusive proof.
3390        OSBORNE v. DISTRICT ATTORNEY’S OFFICE
based on a freestanding claim of actual innocence. Compare
State 

(“conclusively establish Osborne’s
innocence”), with 

(“affirmatively
prove that he is probably innocent”). We therefore afford the
state court’s findings no preclusive effect in determining
whether the evidence in question is sufficiently material to
require disclosure by the State.

                               B

  The State contends that even DNA test results excluding
Osborne as the source of the semen and pubic hair from the
blue condom and the pubic hair from K.G.’s sweater would
not cast sufficient doubt on his conviction to require disclo-
sure of that evidence. In the State’s view, notwithstanding the
prosecution’s reliance on such biological evidence in obtain-
ing Osborne’s conviction, the evidence actually might be
entirely unrelated to the rape and therefore immaterial to
Osborne’s claim of innocence.

   Regarding the semen and pubic hair from the blue condom,
the State acknowledges that the condom was recovered at the
crime scene, that expert testimony was presented at trial com-
paring Osborne’s hair to the pubic hair from the condom and
matching Osborne’s DQ Alpha type to that of the sperm from
the condom, which placed Osborne within 14.7 to 16 percent
of the black population sharing that type, and that in the pros-
ecution’s closing argument it specifically relied on the combi-
nation of this genetic evidence as proof of Osborne’s guilt.
Nonetheless, considering the possibility that more precise
DNA tests might conclusively establish that Osborne did not
use the blue condom, the State now argues that such evidence
was “not the sole basis for finding Osborne guilty” in that
there was circumstantial evidence placing Osborne with Jack-
son on the night in question and K.G. identified Osborne as
the passenger-rapist. Moreover, the State argues, “[o]ther con-
clusions about the donor of the DNA in the sperm and the hair
associated with the condom were plausible and equally recon-
            OSBORNE v. DISTRICT ATTORNEY’S OFFICE           3391
cilable with Osborne’s guilt.” Based on the fact that the con-
dom was recovered by the police more than twenty-four hours
after the assault and, according to the State’s characterization,
“in a semi-secluded area on the outskirts of Anchorage that
was convenient for conducting sexual trysts,” the State now
proposes three possible scenarios for how Osborne could be
guilty despite exculpatory DNA tests. First, the blue condom
could have been discarded at the crime scene by persons unre-
lated to Jackson and Osborne either before K.G.’s attack or
after the attack but before the police searched the area. Sec-
ond, the blue condom could have been used at an earlier time
by another individual, left in Jackson’s car, and discarded or
dropped at the crime scene by Jackson and Osborne. Or third,
trace biological material—particularly the pubic hair found on
the outside of the condom—could have been transferred to the
condom from another surface after its use.

   The State makes a similar argument regarding the pubic
hair that was found on K.G.’s sweater. According to the State,
the fact that K.G. was working as a prostitute, the propensity
hairs have for transference, and the ability of K.G.’s knitted
acrylic sweater to hold a hair all point to the possibility that
it might be anyone’s pubic hair, and not necessarily her
attacker’s. Thus, despite the fact that at Osborne’s trial the
prosecution argued that “we know that the person that did this
had contact with [K.G.’s] sweater” because “[t]here’s a pubic
hair on it,” the State now argues that it has never been estab-
lished that the pubic hair was from one of the assailants and
that it “could have originated from any person who had ever
been near K.G. or from any person who had ever been in
Jackson’s car” prior to the rape.

   [11] On their face, the State’s hypotheticals are not beyond
the bounds of reason. But that is not the applicable federal
standard for determining the materiality of evidence that is
favorable to the accused, notwithstanding the Alaska Court of
Appeals’ interpretation of Alaska law. Cf. State 

-81 (concluding that further DNA testing “would not
3392        OSBORNE v. DISTRICT ATTORNEY’S OFFICE
conclusively establish Osborne’s innocence,” based in part on
the reasoning that the blue condom “might have been coinci-
dentally left in the vicinity by another person before the police
arrived”). Those hypotheticals must be assessed in light of the
entire record. And in that regard, although the State points to
the circumstantial and eyewitness evidence that is supportive
of the prosecution’s case at trial and Osborne’s conviction, the
State fails to point to any evidence in the record that would
affirmatively support its newly imagined alternative theories
of the crime and accordingly rebut its own presentation at trial
regarding the significance of the biological evidence as posi-
tively identifying the real perpetrator. Cf. 

(noting the lack of evidence “in the present record” rebut-
ting the new evidence supporting petitioner’s actual inno-
cence claim). The State’s hypotheticals are formulated based
on nothing more than the very uncertainties that necessarily
arise where new evidence upsets accepted notions of reality
and forces a fundamental reassessment of the factual record.
Even worse, they fail to account for evidence already in the
record that seriously calls into question whether the State’s
hypotheticals might have any basis in reality. Cf. 

(“This should be a matter for the trier of fact to consider in
the first instance, but we can note a line of argument that
could refute the State’s position.”).

   The trial record reveals that the attack occurred down a ser-
vice road, which was located in an isolated area on the out-
skirts of Anchorage off Point Woronzof Drive and near
Earthquake Park. Although the State characterizes it as a
“semi-secluded” location that was “convenient for conducting
sexual trysts,” which is apparently why the perpetrators chose
the location, there does not appear to be any evidence in the
record indicating that the location was ever used for such a
purpose by any other persons on that night or the following
day, or for that matter any other day. To the contrary, the evi-
dence presented at trial seems to indicate that no one else had
been down that service road, either by car or on foot, around
the time of the crime. There was already snow on the ground
            OSBORNE v. DISTRICT ATTORNEY’S OFFICE          3393
before the attack, and a layer of new snow fell by the next
morning. Judging by the tracks in the snow, only a few known
individuals visited the area at the end of the service road both
on the night of the assault and the next day. Even though the
police did not arrive until twenty-four hours after the attack,
they found only a few sets of footprints and tire tracks. The
police made castings of some of the show prints in the snow,
but those were matched to K.G.’s rescuers and their neighbor
and were therefore eliminated as suspect prints. Similarly, the
trial record indicates the police found only three sets of tire
tracks, all of which were later matched to known vehicles—
those owned by Jackson, K.G.’s rescuers, and their neighbor.
Because of the old and new snow layers, the police were even
able to distinguish each of the three sets of tracks by the time
periods in which they were made. The tracks from Jackson’s
car were in a lower layer of snow, while the two other sets
were fresh. This was due to the fact that K.G. had walked to
the main road before being picked up, and her rescuers and
their neighbor did not visit the crime scene until the following
day after the new snow had fallen. Significantly, the trial
record does not indicate the existence of any other tire tracks
or shoe prints that might have indicated that some unknown,
innocent person visited the scene either before or after the
attack.

   Additional record evidence also tends to contradict the
notion that the blue condom could be attributable to someone
other than Jackson’s accomplice. The trial record indicates
that the police recovered no condoms from Jackson’s car, they
recovered only one condom and part of a blue foil condom
wrapper from the crime scene, and the color and size of the
blue condom and the color of the wrapper were the same as
that of the condom and wrapper that K.G. had with her that
night. K.G. also testified that she thought the condom was the
same one that she had with her and that Jackson’s passenger
wore during the rape. The police found the condom “at point
C in the snow” right beside some blood and part of the blue
and gold foil wrapper matching the condom K.G. was carry-
3394        OSBORNE v. DISTRICT ATTORNEY’S OFFICE
ing, and it was also very near the spent shell casing from
Jackson’s gun, K.G.’s bloody pants, the disturbed berm of
snow in which K.G. had been partially buried, and the tire
tracks from Jackson’s car, probably on the passenger side
where Jackson’s accomplice would have been. Thus, the con-
dom was found not merely in the vicinity of the crime scene
but in the exact location where K.G. was attacked. Regarding
the timing, although the police did not search the crime scene
until twenty-four hours after the rape, K.G.’s rescuers
reported seeing the condom lying in the snow when they vis-
ited the crime scene around noon on the day after the attack.
Thus, the time frame in which the condom could have been
left there was far narrower than the State suggests. Finally, a
crime lab technician testified that when he received the con-
dom, which by then had been packaged in a closed plastic
cup, the condom was rolled down and there was still a sub-
stantial amount of semen inside that had not yet dried. Taken
together, these facts are far more consistent with the prosecu-
tion’s trial narrative—that Jackson’s passenger used K.G.’s
condom during the rape, he removed and discarded it at the
scene, and it remained undisturbed in the snow until it was
recovered by the police—than any of the State’s newly pro-
posed alternative explanations for the condom’s origin.

   The State also fails to recognize that further DNA testing
is alone capable of establishing the supposedly missing link
between the condom and K.G.’s rape. The potential probative
value of the semen and pubic hair from the blue condom and
the pubic hair from K.G.’s sweater must be considered collec-
tively, not in isolation as the State has considered the evi-
dence. See 

(explaining that materiality
is defined “in terms of suppressed evidence considered collec-
tively, not item by item”); 

(“Schlup
makes plain that the habeas court must consider all the evi-
dence, old and new, incriminating and exculpatory” (internal
quotation marks omitted)); see, e.g., 

(“If consid-
ered in isolation, a reasonable jury might well disregard it. In
combination, however, . . . the evidence . . . likely would rein-
            OSBORNE v. DISTRICT ATTORNEY’S OFFICE          3395
force other doubts as to House’s guilt.”). As the State itself
should expect given the prosecutor’s trial presentation and its
continued belief in Osborne’s guilt, further DNA testing could
establish a genetic match between the semen and pubic hair
on the condom and the pubic hair on K.G.’s sweater, proving
that the user of the condom was also in such contact with
K.G. as to transfer a pubic hair to her sweater. As the prosecu-
tor argued at trial, because the sweater was spread out under-
neath K.G. during the rape, the most likely scenario is that the
hair was transferred by Jackson’s accomplice at that time.

   Moreover, further DNA testing might even directly link the
blue condom to K.G. herself. Although we have mainly
focused on the semen and hair evidence that we know to exist,
the trial record reveals that epithelial cells were found on the
outside of the condom. Based on this evidence, the prosecu-
tion argued at trial that it indicates how the condom was used
because epithelial cells come from only the inside of the
mouth, rectum or vagina. Further DNA testing might be able
to go one step further and genetically match any such trace
material from the condom to K.G., conclusively establishing
that the condom was used in K.G.’s rape and disproving any
of the State’s new hypotheses.

   New evidence favorable to Osborne could also lead to new
lines of investigation and additional new evidence. See People
v. Garcia, 

, 551 (Cal. Ct. App. 1993); see
also 

(“[T]he reviewing court may
consider directly any adverse effect that the prosecutor’s fail-
ure to respond might have had on the preparation or presenta-
tion of the defendant’s case.”). Based on the State’s obstinate
denial that Osborne might be innocent even if DNA test
results exclude him as a source of the biological evidence
used to convict him, the State would seem intent on develop-
ing a case to continue to fight Osborne’s claim of innocence.
But such an investigation might instead lead in the opposite
direction and further solidify Osborne’s case for innocence.
As Osborne contends, if the STR DNA test results exclude
3396        OSBORNE v. DISTRICT ATTORNEY’S OFFICE
him, those results could then be placed into the state and
national DNA databank system, which did not exist when
Osborne was tried, and possibly identify the real perpetrator.
See State 

n.14 (noting that Alaska main-
tains a DNA registry).

   [12] In summary, the State’s proposed hypotheticals for
reconciling exculpatory DNA tests with Osborne’s guilt are so
inconsistent with and improbable in light of the evidence in
the trial record that they cannot negate the materiality of fur-
ther DNA testing to possible post-conviction relief. Cf.

(holding that, although there was
not “conclusive exoneration,” the petitioner satisfied the Sch-
lup actual innocence standard). “In light of the obvious excul-
patory potential of semen evidence in a sexual assault case,”

n.2, and given the evidentiary record
in this sexual assault case and the unique circumstances of
this crime, we have no difficulty concluding that DNA tests
favorable to Osborne would have extraordinary exculpatory
potential and would be material to proving his actual inno-
cence. At the very least, exculpatory DNA tests would entitle
Osborne to an evidentiary hearing on his actual innocence
claim in order to more fully develop the factual record and
reconcile any conflicting evidence. See Cooper v. Woodford,

, 1123-24 (9th Cir. 2004) (en banc); cf. 

(addressing the merits of a Schlup gateway
claim “based on a fully developed record”).

                               C

   Thus far we have limited our analysis to the evidence in the
trial record and the evidence that might be discovered if the
State were forced to allow it to come to light. But the State
also points to Osborne’s written confession in his application
for parole and his oral confession at his subsequent parole
hearing as foreclosing his right to obtain post-conviction
access.
             OSBORNE v. DISTRICT ATTORNEY’S OFFICE             3397
   The confessions are certainly relevant to our inquiry.
Because Osborne’s ultimate claim for post-conviction relief
will be actual innocence rather than mere trial error, all new
evidence may be considered in assessing the potential materi-
ality of further DNA testing. See 

(“all
the evidence, old and new, incriminating and exculpatory,
without regard to . . . admissibility . . . at trial” (internal quo-
tation marks omitted)). No doubt, that includes Osborne’s
written and oral confessions to the state parole board. See
State 

-79.

   [13] We disagree, however, that the confessions foreclose
Osborne’s right to obtain post-conviction access to evidence.
The same rule that allows us to consider the probative value
of the confessions requires that we do so in light of exculpa-
tory DNA tests and all the rest of the new and old evidence
in this case. Thus, the question before us is not how much
weight we should afford Osborne’s confessions standing
alone, but how they might be squared with exculpatory DNA
tests and the remainder of the evidentiary record. See Gods-
chalk v. Montgomery County Dist. Attorney’s Office, 177 F.
Supp. 2d 366, 370 (E.D. Pa. 2001) (“While plaintiff’s detailed
confessions to the rapes are powerful inculpatory evidence, so
to any DNA testing that would exclude plaintiff as the source
of the genetic material taken from the victims would be pow-
erful exculpatory evidence.”). As we have already discussed,
further DNA testing will be highly probative of Osborne’s
guilt or innocence given the facts as we know them. We can
therefore expect to see one of two possible scenarios unfold:
Either the confessions will be proven accurate by test results
proving Osborne was in fact the passenger-rapist and his case
will proceed no further, or the test results will exclude him as
the source of the biological material, in which case serious
questions will be raised about the validity of his confessions
and whether, as Osborne now claims, he was motivated to
confess falsely as the most expeditious means available to
obtain release from prison. Cf. Baylor v. Estelle, 

, 1323-25 (9th Cir. 1996) (holding that despite the defen-
3398         OSBORNE v. DISTRICT ATTORNEY’S OFFICE
dant’s confession, trial counsel’s ineffective assistance in fail-
ing to follow up on potentially exculpatory semen evidence in
a sexual assault case was prejudicial because evidence exclud-
ing the defendant as the semen donor “would necessarily have
raised reasonable doubt about the validity of his confession”).

   [14] Accordingly, we decline to hold that Osborne’s con-
fession during parole proceedings necessarily trumps the
materiality of physical evidence or the right to obtain post-
conviction access to evidence. Such a rule would ignore the
emerging reality of wrongful convictions based on false con-
fessions and the capability of DNA testing to reveal the objec-
tive truth and exonerate the innocent.

                                D

   The State finally contends that the district court erred in
finding that further DNA testing “can be easily performed
without cost or prejudice to the [State].” Osborne, 445 F.
Supp. 2d at 1081. In the State’s view, prejudice is inherent in
the granting of post-conviction access to evidence because it
erodes the important value of finality in the criminal justice
system. We disagree.

   Although finality is undoubtedly an important consider-
ation, it is not such an immovable force as to override the due
process interests presently at stake. If Osborne already had in
hand the exculpatory evidence he seeks and filed a habeas
petition stating a valid claim for relief, there would be no
question that his petition must be heard despite finality con-
siderations. “The federal writ of habeas corpus overrides all
[such] considerations, essential as they are to the rule of law,
when a petitioner raises a meritorious constitutional claim in
a proper manner in a habeas petition.” McCleskey v. Zant, 

, 492-93 (1991). Even a procedural default would not
necessarily foreclose Osborne’s claim given the availability of
a Schlup gateway claim of actual innocence. See, e.g.,

. Though no doubt eroding finality,
            OSBORNE v. DISTRICT ATTORNEY’S OFFICE           3399
such an exception “serves as ‘an additional safeguard against
compelling an innocent man to suffer an unconstitutional loss
of liberty,’ guaranteeing that the ends of justice will be served
in full.” 

(quoting Stone v. Powell,

, 491-92 n.31 (1976)).

   The State’s conception of finality would reverse these pri-
orities. The evidence in question can be produced easily and
without cost to the State and, if favorable to Osborne, would
be strong evidence in support of post-conviction relief. None-
theless, the State seeks to foreclose such relief by its simple
refusal to open the evidence locker. We rejected a similar tac-
tic in 

, and we reject it again here.

   The State supports its position with the argument that the
circumstantial and eyewitness evidence in this case is also
strong evidence of Osborne’s guilt, and thus granting access
is not likely to “further the truth seeking function of our crim-
inal justice system.” As recent history has shown, however,
DNA evidence has the capability of refuting otherwise irrefut-
able inculpatory evidence, and as we have already established
this case is no exception.

   If the inculpatory evidence has been correctly interpreted,
further DNA testing will confirm that Osborne is guilty as
charged and convicted. But it remains a very real possibility
that further DNA testing will be exculpatory and may even
lead to Osborne’s exoneration. In the former case, the State
will have lost nothing; indeed, it will gain even more defini-
tive proof of Osborne’s guilt and will be relieved of the bur-
den of further post-conviction litigation. In the latter case,
however, Osborne will obviously gain a great deal, as will the
State, whose paramount interests are in seeking justice, not
obtaining convictions at all costs, and which will then have
strong evidence for use in catching and punishing the real per-
petrator. Importantly, the State is prejudiced in neither case,
and the truth-seeking function of the criminal justice system
is furthered in either case.
3400           OSBORNE v. DISTRICT ATTORNEY’S OFFICE
                                    V

   [15] In Thomas, we granted a prisoner’s request for post-
conviction DNA testing to establish a gateway claim of actual
innocence “[i]n light of the obvious exculpatory potential of
semen evidence in a sexual assault 

n.2.
This sexual assault case is no exception. We therefore agree
with the district court and hold that Osborne’s right to due
process of law prohibits the State from denying him reason-
able access to biological evidence for the purpose of further
DNA testing, where that biological evidence was used to
secure his conviction, the DNA testing is to be conducted
using methods that were unavailable at the time of trial and
are far more precise than the methods that were then avail-
able, such methods are capable of conclusively determining
whether Osborne is the source of the genetic material, the
testing can be conducted without cost or prejudice to the
State, and the evidence is material to available forms of post-
conviction relief.4

   In so holding, however, we do not purport to set the stan-
dards by which all future cases must be judged. We are pre-
sented with a certain set of circumstances presenting a merito-
rious case for disclosure, and our analysis and holding are
addressed to those circumstances only. Despite the manner in
which the parties have presented the issues, such questions as
whether the scope of the right of post-conviction access
should be broader or flexible to accommodate different cir-
cumstances, whether the materiality standard for post-
conviction access-to-evidence claims should be less stringent
or defined in a different manner, and whether prisoners with
  4
    Given our holding, we need not reach Osborne’s alternative arguments
that the State’s denial of access to potentially exculpatory DNA evidence
is effectively a denial of meaningful access to courts in violation of the
First and Fourteenth Amendments, see Christopher v. Harbury, 

, 412-22 (2002), or that it violates his due process right to effectively
pursue parole and executive clemency, see Harvey 

(Luttig, J., respecting the denial of rehearing en banc).
           OSBORNE v. DISTRICT ATTORNEY’S OFFICE       3401
a less compelling case might also be entitled to post-
conviction access, all are questions that we need not answer
and do not purport to answer in deciding this case. We leave
them for another day.

  AFFIRMED.