FOR PUBLICATION
 UNITED STATES COURT OF APPEALS
      FOR THE NINTH CIRCUIT

LARRY BERMAN,                                No. 05-16820
                 Plaintiff-Appellant,
                  v.                           D.C. No.
                                            CV-04-02699-DFL
CENTRAL   INTELLIGENCE AGENCY,
                                               OPINION
                Defendant-Appellee.
                                        
      Appeal from the United States District Court
         for the Eastern District of California
        David F. Levi, District Judge, Presiding

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

                   Filed September 4, 2007

   Before: David R. Thompson, Pamela Ann Rymer and
           Raymond C. Fisher, Circuit Judges.

                   Opinion by Judge Fisher




                             11337
11340                   BERMAN v. CIA


                         COUNSEL

Thomas R. Burke and Duffy Carolan (argued), Davis Wright
Tremaine LLP, San Francisco, California; and Meredith
Fuchs, The National Security Archive, Washington, D.C., for
the plaintiff-appellant.

Peter D. Keisler, McGregor W. Scott, Mark B. Stern (argued)
and Alisa B. Klein, U.S. Department of Justice, Washington,
D.C., for the defendant-appellee.

Matthew W.S. Estes, Washington, D.C., for the amici curiae
The American Historical Association, et al.


                          OPINION

FISHER, Circuit Judge:

   For nearly half a century, the CIA has each day sent the
President a highly classified summary of the most important
and timely intelligence relating to this country’s national
defense and foreign policy priorities. We must decide in this
case whether two of these reports — known as the President’s
Daily Brief (PDB) — from the administration of President
Lyndon B. Johnson are exempt from disclosure under the
Freedom of Information Act (FOIA), 5 U.S.C. § 552. We hold
that the CIA has provided ample justification that the disclo-
sure of the two PDBs would reveal protected intelligence
sources and methods, and thus these PDBs are protected by
FOIA exemption 3 and the National Security Act (NSA), 50
U.S.C. §§ 403-1(i)(1), 403g.

                               I.

  The practice of specialized presidential intelligence briefing
dates back to the administration of President John F.
                        BERMAN v. CIA                      11341
Kennedy. After taking office, President Kennedy asked the
CIA to produce a special briefing that succinctly summarized
recently collected intelligence information that would be of
interest to the President and his senior advisors. That briefing,
which was then called the President’s Intelligence Checklist
(PICL), became an important medium of communication
between the leadership of the CIA and the White House.
When President Johnson took office, the PICL’s format was
modified to suit his particular tastes, and was renamed the
President’s Daily Brief. The PDBs of that era reported on
international developments based on intelligence sources that
included satellite photographs, signal intercepts, individual
recruits, Department of State communications, published
news accounts and other publicly available information.
Because the PDBs were high-level intelligence documents,
they were then and still are classified documents that are
available only to the President and his senior advisors.

   Over the years, a handful of the more than 13,500 existing
PICLs and PDBs have made their way into the public domain,
either deliberately or by mistake. Ten redacted PICLs from
the Kennedy administration were released pursuant to the
President John F. Kennedy Assassination Records Collection
Act of 1992. See 44 U.S.C. § 2107 note. Two more recent
PDBs were released as a part of the Final Report of the
National Commission on Terrorist Attacks on the United
States, commonly known as the 9/11 Report. These PDBs
were declassified after the Director of Central Intelligence
determined that the public interest in disclosure outweighed
the potential damage to national security that could result
from disclosure. See Exec. Order No. 12,958, 68 Fed. Reg.
15315, § 3.1(b) (as amended by Exec. Order No. 13,292).

  At least 15 redacted PDBs from the Johnson administration
have also been released. These PDBs illustrate the format and
content that was common during that period. They were pro-
duced in a two-column format with particular countries listed
on the left and one or two paragraphs about recent events in
11342                   BERMAN v. CIA
each country on the right. The content of these PDBs are gen-
erally factual, although in some cases the author provides pre-
dictions about where current events might lead. The tone is
generally informal.

   Larry Berman, a political science professor at the Univer-
sity of California, Davis, filed a FOIA request seeking two
Johnson-era PDBs: from August 6, 1965 and from April 2,
1968. The CIA denied his request, asserting FOIA exemptions
for classified national security information (exemption 1); for
protected intelligence sources and methods (exemption 3);
and for privileged communications (exemption 5). See 5
U.S.C. § 552(b)(1), (b)(3), (b)(5). After his administrative
appeal was denied, Berman filed a declaratory judgment
action in the Eastern District of California seeking disclosure.

   In the district court proceedings, the CIA supported its
asserted exemptions with the 39-page declaration and three-
page supplemental declaration of CIA information review
officer Terry Buroker. In his declarations, Buroker asserts that
the PDBs “must be withheld in their entirety, as no reasonably
segregable, non-exempt portions of the documents exist.”
Buroker provides two related factual bases for the claimed
necessity of keeping the PDBs secret.

   First, Buroker describes the general content and function of
PDBs. He explains that during the Johnson administration,
PDBs were used to synthesize, in a few pages, the most
recently gathered and critical intelligence information the CIA
possessed. Because of their condensed format, the Johnson
PDBs contained only that information that leadership within
the CIA believed would be most important to the President
and his senior advisors. Buroker explains that PDBs served as
a starting point for high level discussions regarding intelli-
gence and national security between the President and the
CIA. As a result, the PDBs themselves reflect one side of this
ongoing dialogue. According to Buroker, the Johnson PDBs
include sensitive information such as: “a) undisseminated raw
                        BERMAN v. CIA                    11343
operational information, sometimes including true names of
sources and/or cryptonyms, b) sensitive operational informa-
tion added to the document by the Directorate of Operations
after the Directorate of Intelligence has written or edited the
material in the PDB, c) information restricted at the very
highest levels of human and technical source intelligence
gathering, d) information from covert technical operations,
and e) information from specifically developed or acquired
CIA-only methods.”

   Second, Buroker discusses why the specific PDBs
requested in this case would result in harm to the CIA’s intel-
ligence gathering interests. Buroker states that the specific
PDBs Berman requests “contain explicit references to infor-
mation provided by foreign officials as well as other informa-
tion that may incorporate information from foreign liaison
relationships,” including foreign governments and foreign
intelligence services. The PDBs also “contain references to
intelligence obtained from individual human sources and from
confidential liaison relationships.” Buroker warns that such
information was provided “only upon a guarantee of absolute
secrecy,” and disclosure of the requested PDBs “would tend
to reveal the identities of intelligence sources.” This could
lead to severe harms to the sources of the information, includ-
ing “embarrassment, political ruin, retribution . . . imprison-
ment, torture or even death of the source or the source’s
family and friends.” Moreover, Buroker states that disclosure
of the requested PDBs “would disclose specific intelligence
methods, including technical collection methods.” For all of
these reasons, disclosure “reasonably could be expected to
cause exceptionally grave damage to the national security of
the United States.”

   The district court granted summary judgment in favor of
the CIA, holding that the CIA had made an adequate showing
that the documents were shielded from disclosure by exemp-
tions 3 and 5. See Berman v. CIA, 

 (E.D.
Cal. 2005). This timely appeal followed.
11344                       BERMAN v. CIA
                                    II.

   The district court’s grant of summary judgment in a FOIA
case is reviewed under a two-step test. See Lion Raisins v.
U.S. Dep’t of Agric., 

, 1078 (9th Cir. 2004).
First, we ask whether the district court had an adequate fac-
tual basis for its decision. We review the district court’s deter-
mination that a particular set of documents (here the Buroker
declarations) provided an adequate factual basis de novo. Sec-
ond, we ask whether the district court’s decision regarding
applicability of FOIA’s exemptions was correct. If the district
court’s determination turns mainly on findings of fact, we
review for clear error. Id. However, where as here the district
court’s determination turned on its interpretation of the law,
we review de novo. See generally Schiffer v. FBI, 

, 1409 (9th Cir. 1996) (“Although any factual conclu-
sions that place a document within a stated exemption of
FOIA are reviewed under a clearly erroneous standard, the
question of whether a document fits within one of FOIA’s
prescribed exemptions is one of law, upon which the district
court is entitled to no deference” (citation and internal quota-
tion marks omitted)).

                                   III.

                                    A.

   [1] FOIA exemption 3 permits government agencies to
maintain the secrecy of information that is “specifically
exempted from disclosure by [certain] statute[s] . . . .” See 5
U.S.C. § 552(b)(3). The National Security Act is such a stat-
ute. See CIA v. Sims, 

, 167 (1985). That statute
instructs the Director of National Intelligence to “protect
intelligence sources and methods from unauthorized disclo-
sure.” 50 U.S.C. § 403-1(i)(1); see also 50 U.S.C. § 403g.1
  1
    The statute formerly referred to the Director of Central Intelligence.
See Sims, 471 U.S. at 167. The change in titles and responsibilities has no
impact on this case. See Wolf v. CIA, 

, 377 n.6 (D.C. Cir.
2007).
                         BERMAN v. CIA                     11345
   [2] The NSA provides the Director with “very broad
authority to protect all sources of intelligence information
from disclosure.” Sims, 471 U.S. at 168-69. Because of this
“sweeping power,” id. at 169, courts are required to give
“great deference” to the CIA’s assertion that a particular dis-
closure could reveal intelligence sources or methods, id. at
179. The term “sources” is to be broadly construed and
encompasses not only “secret agents,” but instead reaches all
sources of information the CIA relies upon, including publicly
available information. Id. at 170-71.

   [3] We have acknowledged that after Sims, there exists “a
near-blanket FOIA exemption” for CIA records. Hunt v. CIA,

, 1120 (9th Cir. 1992). Indeed, Sims leaves
courts “only a short step from exempting all CIA records from
FOIA.” Id. (internal quotation marks and alteration omitted).
Concerned that this broad reading of CIA authority might be
contrary to congressional intent, we have invited Congress to
“take the necessary legislative action to rectify” that disparity.
Id.; see also Minier v. CIA, 

, 804 (9th Cir. 1996)
(“Only Congress can override the plain language of [the
NSA].”). Congress, however, has to date left the NSA materi-
ally unaltered and so we must continue to afford the CIA
broad deference. Nonetheless, just as Congress has not
reduced the CIA’s authority under the NSA, neither has it
expanded the CIA’s protection from FOIA from a “near-
blanket” exemption to a blanket exemption. We therefore
continue to conduct some meaningful — albeit restrained —
review of the CIA’s assertions.

   [4] The CIA bears the burden of proving the applicability
of the exemption. Minier, 88 F.3d at 800. Although the CIA’s
reasons are entitled to deference, the CIA’s declarations must
still “describe the justifications for nondisclosure with reason-
ably specific detail, demonstrate that the information withheld
logically falls within the claimed exemptions, and show that
the justifications are not controverted by contrary evidence in
the record or by evidence of CIA bad faith.” Hunt, 

                      BERMAN v. CIA
at 1119. The CIA must do more than show simply that it has
acted in good faith. See Wiener v. FBI, 

, 983
n.19 (9th Cir. 1991).

                                   B.

   Buroker asserts that exemption 3 applies because the PDBs
could “expose the existence of specific intelligence sources
and methods.”2 He explains that each of the requested PDBs
contains “information specifically stating sensitive sources or
methods of collection” and would reveal “substantial informa-
tion about its provenance to an educated reader.” Buroker
adds that “each edition of the PDB is a piece of a ‘mosaic’ of
information reflecting the most sensitive, as well as the mun-
dane, intelligence sources and methods employed by the CIA
and the Intelligence Community over time.”

   In Wiener, we held that a CIA affidavit was inadequate to
support the CIA’s invocation of exemption 3 because it “fail-
[ed] to discuss the facts or reasoning upon which [the declar-
ant] based his conclusion.” Wiener, 943 F.2d at 983. This was
problematic because it denied the plaintiff the opportunity to
contest the CIA’s conclusions, and thus distorted the adver-
sary process. See id. The CIA affidavit we rejected stated
without justification that “disclosure of [the withheld] por-
tions reasonably could be expected to lead to identification of
the source of the information.” Id. (alteration in original,
internal quotation marks omitted).

  [5] Unlike the affidavit in Wiener, Buroker’s declaration
provides the facts and reasoning upon which his conclusion is
based. He explains that each PDB contains the most sensitive
and important intelligence information available to the CIA
  2
   Because we hold that the PDBs are protected under exemption 3, we
do not decide whether the CIA’s claims that they are also protected under
exemption 1 and exemption 5 are valid. See Minier, 88 F.3d at 800 n.5;
Hunt, 981 F.2d at 1118.
                        BERMAN v. CIA                     11347
on the day it is released and that the requested PDBs specifi-
cally identify intelligence sources and methods. He also
explains how PDBs convey important contextual information
that could reveal sources or methods even if the PDBs were
produced in redacted form. Because they are released on a
daily basis and typically contain only the most current infor-
mation fresh from the field, PDBs reveal when particular
intelligence information became available to the CIA. This is
important, as Buroker observes, because sophisticated foreign
intelligence services might use that information to “reliably
infer that a human source for information contained in the
PDB is most likely one of a very few number of individuals
with access to the subject information, and that the source
must have provided the information very close in time to
when it was reported in the PDB.” An educated observer
could also determine other sensitive information, such as what
intelligence was most important to the President and to senior
officials within the CIA at a particular time. See Sims, 471
U.S. at 176-77 (“A foreign government can learn a great deal
about the Agency’s activities by knowing the . . . sources of
information that interest the Agency.”). Furthermore, Buroker
explains, the release of PDBs would diminish the CIA’s abil-
ity to assure current intelligence sources that their identities
will be kept secret in the future.

   [6] Finally, in addition to the Buroker affidavit, Berman has
access to several other Johnson PDBs that undisputedly con-
tain a similar form and content to the ones he requests. He
also has access to volumes of other information regarding
intelligence and foreign policy during the Johnson administra-
tion, including National Security Files available through the
Johnson Presidential Library. Access to these documents has
enabled him specifically to contest the CIA’s claim that the
PDBs would divulge protected sources and methods. Berman
therefore has sufficient facts at his disposal for the adversary
process to function properly and thus he has been given “a
meaningful opportunity to contest, and the district court an
adequate foundation to review, the soundness of the withhold-
11348                   BERMAN v. CIA
ing.” Weiner, 943 F.3d at 977 (citation and internal quotation
marks omitted). Because of the broad deference we are to
give the CIA under Sims, and because judges are poorly posi-
tioned to evaluate the sufficiency of the CIA’s intelligence
claims, see Sims, 471 U.S. at 176, 178, we doubt that the
CIA’s provision of a more detailed declaration would enable
Berman to argue more effectively for their release. See
Weiner, 943 F.2d at 983 (stating that the CIA affidavit was
sufficient with regard to one withholding because “[n]o fur-
ther disclosure would have enabled Weiner to argue for their
release”).

   [7] Berman argues nonetheless that the CIA should be
required to provide even greater detail regarding the content
of the requested PDBs and how that content is tied to the
harms the CIA fears. We are satisfied, however, that the
Buroker declaration strikes the appropriate balance between
justifying the applicability of the exemption with sufficient
specificity to permit Berman meaningfully to challenge it and
the CIA’s need to avoid providing a description that is so spe-
cific that it risks revealing protected sources and methods. As
Sims observed, “[i]t is conceivable that the mere explanation
of why information must be withheld can convey valuable
information to a foreign intelligence agency.” 471 U.S. at
179; see also Church of Scientology of Cal. v. U.S. Dep’t of
Army, 

, 742 (9th Cir. 1979) (stating that in
asserting a FOIA exemption “the government need not spec-
ify its objections in such detail as to compromise the secrecy
of the information”). We were mindful of this concern in
Hunt, where we permitted the CIA to respond to a FOIA
request by providing a so-called “Glomar response” in which
the CIA refused to confirm or deny the existence of records
pertaining to a foreign national. 981 F.2d at 1117. We did not
require the CIA to identify particular harms that would occur
if the documents were disclosed. Instead, we were satisfied
with the CIA’s statements that a more specific response might
allow foreign intelligence agents to “determine the contours
and gaps of CIA intelligence operations and make informed
                           BERMAN v. CIA                         11349
judgments as to the identities of probable sources and targets
in other countries,” and that disclosure might prove to be a
disincentive to future sources providing assistance the CIA.
Id. at 1119. We cited with approval Gardels v. CIA, 

 (D.C. Cir. 1982), in which the CIA justified a Glomar
response by averring that disclosure might allow foreign oper-
atives to discover the identities of covert CIA sources and the
CIA research interests. Hunt, 981 F.2d at 1119-20 (citing
Gardels, 689 F.2d at 1103-04). It was therefore unnecessary
for the affidavits in that case to “mention harms to particular
individuals.” Id. The CIA’s declarations here are comparable
in specificity to those in Hunt and Gardels. We therefore hold
that they are sufficient.3

   [8] Berman next argues that Buroker’s declarations are
insufficient because Buroker states only that disclosure of the
PDBs “could” reveal sources and methods, rather than stating
definitively that the PDBs would divulge such protected mat-
ter. But Buroker’s declaration speaks in more certain terms
than Berman suggests, in stating that the disclosure of the
requested PDBs “would disclose specific intelligence meth-
ods, including technical collection methods,” that disclosure
“would tend to reveal the identities of intelligence sources,”
and that “[e]ach of the Requested PDBs contains information
specifically stating sensitive sources or methods of collec-
tion.” (Emphases added). More fundamentally, the CIA
Director need not demonstrate to a certainty that disclosure
will result in intelligence sources or methods being revealed.
Instead, the NSA entrusts the Director with the discretion to
determine that documents should remain secret because the
  3
   In arguing that the CIA is required to disclose even more details
regarding the contents of the requested PDBs, Berman cites precedent
from this court and elsewhere regarding the appropriate standard where
exemption 1 is at issue and where the CIA is not involved. But given the
especially broad deference that we must accord the CIA under Sims, stan-
dards established under exemption 1 cannot be uncritically imported into
the exemption 3 context. See Fitzgibbon v. CIA, 

, 764 (D.C.
Cir. 1990).
11350                   BERMAN v. CIA
substantial risk that sources and methods will be compro-
mised outweighs the public interest in disclosure. In Sims, the
Supreme Court deferred to the CIA’s conclusion that disclo-
sure of the requested information “posed an unacceptable risk
of revealing protected ‘intelligence sources.’ ” 471 U.S. at
179; see also Wolf, 473 F.3d at 377 (“[I]nformation is exempt
under [the NSA] if the Agency demonstrates that an answer
to the query can reasonably be expected to lead to unautho-
rized disclosure.” (emphasis added) (citation and quotation
marks omitted)). We must therefore defer to the CIA’s deter-
mination that disclosure would run the unacceptable risk that
sources or methods would be revealed. Buroker has asserted
that such risks are present here, not only with regard to revela-
tion of sources and methods, but also in his warning that dis-
closure “reasonably could be expected to cause exceptionally
grave damage to the national security of the United States.”
Because the CIA is better situated to gauge the national secur-
ity implications of disclosure, see Sims, 471 U.S. at 179, we
defer to its judgment.

   [9] Berman also objects to the CIA’s reliance on the “mo-
saic theory,” which Buroker explains as the concept that
“[w]hile some information in specific PDBs may appear
harmless to disclose when read in isolation, such information
may be very valuable as part of a ‘mosaic’ of information
gleaned from various sources, including multiple PDBs pre-
pared over time.” The Supreme Court endorsed the mosaic
theory in Sims, commenting that

    [B]its and pieces of data may aid in piecing together
    bits of other information even when the individual
    piece is not of obvious importance in itself. Thus,
    [what] may seem trivial to the uninformed, may
    appear of great moment to one who has a broad view
    of the scene and may put the questioned item of
    information in its proper context. Accordingly, the
    Director, in exercising his authority under [the
    NSA], has the power to withhold superficially innoc-
                         BERMAN v. CIA                     11351
    uous information on the ground that it might enable
    an observer to discover the identity of an intelligence
    source.

471 U.S. at 178 (citations and internal quotation marks omit-
ted; second alteration in original). We permitted the CIA to
rely upon the mosaic theory in Hunt, when we accepted the
CIA’s argument that “disclosure . . . must not be viewed in
isolation but rather as one tile in a mosaic of intelligence gath-
ering.” 981 F.2d at 1119; accord Fitzgibbon, 911 F.2d at 763;
see also Kasza v. Browner, 

, 1166 (9th Cir.
1998) (endorsing the mosaic theory in context of the govern-
ment’s assertion of the state secrets privilege against civil dis-
covery requests). To be sure, the CIA’s invocation of the
mosaic theory does not excuse it from meeting its burden of
proof or from its obligation to provide a reasonably specific
explanation of why the exemption applies. But nothing pre-
vents the CIA from relying on the common sense premise that
the impact of disclosing protected documents must be evalu-
ated not only based upon the information appearing within the
four corners of the document, but also with regard to what
secrets the document could divulge when viewed in light of
other information available to interested observers.

   We are to consider the sufficiency of the CIA’s declara-
tions in light of contrary evidence adduced by Berman. See
Hunt, 981 F.2d at 1119. Berman relies upon the several PDBs
that have been publicly disclosed to argue that there is nothing
about PDBs generally that necessitates secrecy and thus the
CIA was required to provide more detail regarding the PDBs
requested in this case. Buroker’s declaration explains, how-
ever, that each additionally disclosed PDB would provide
more clues regarding intelligence sources and methods, as
well as the ongoing dialogue between the President and the
CIA. If multiple PDBs within a short time period were avail-
able, then an informed observer would be able to trace pre-
cisely when new information became available and, in the
case of multiple entries on the same country, what types of
11352                   BERMAN v. CIA
sources were employed. The CIA’s explanations are entitled
to deference and, in light of the overall showing made in
Buroker’s declarations, provide a sufficient factual basis for
the district court’s decision. Cf. Sims, 471 U.S. at 180-81
(rejecting the argument that the CIA was “somehow estop-
ped” from withholding the identities of some intelligence
sources because it had already disclosed others).

   Berman also contends that much of the information con-
tained in PDBs is similar if not identical to information set
forth in publicly available Central Intelligence Bulletins
(CIBs). Like PDBs, CIBs are brief summaries of intelligence
information distributed to senior members of the executive
branch. Buroker explains, however, that unlike PDBs, CIBs
are not developed exclusively for the President; they are cir-
culated throughout the government, including to policy,
security and military officials. CIBs do not contain raw intelli-
gence or direct source information and are written at a greater
level of generality than PDBs The distinction drawn by
Buroker is supported by the record. Although the CIBs in
some cases contain similar text to contemporaneous PDBs,
the majority of the information presented in the publicly
available PDBs does not overlap with the same-day CIBs.
Even if we assume that all the information in a particular PDB
eventually makes its way into the CIBs — an assumption the
record does not support — revelation of the PDBs would still
divulge when that information became available and therefore
potentially signal what sources or methods were in play. It
might also reveal what information was of primary interest to
the President at a given time. See Sims, 471 U.S. at 178
(“Foreign intelligence services have an interest in knowing
what is being studied and researched by our agencies dealing
with national security . . . .”).

   [10] Accordingly, we hold that the Buroker affidavits and
other evidence in the record provided an adequate factual
basis for the district court’s decision. The CIA did not need
to provide a more specific declaration.
                        BERMAN v. CIA                      11353
                               C.

   [11] Because Buroker has stated explicitly that the
requested PDBs would reveal protected sources and methods
if disclosed, and because his declarations adequately support
that assertion, exemption 3 applies. If Berman were seeking
more recent PDBs there would be little room for dispute that
they would reveal highly sensitive intelligence sources and
methods. Berman’s request therefore relies heavily upon the
assumption that the passage of time — around 40 years — has
eroded any support for the assertion that the PDBs contain
information about sources or methods that has not already
been revealed.

   Buroker explains, however, that the passage of time has not
vitiated the CIA’s interest in maintaining the secrecy of the
requested PDBs. In this regard, some of Buroker’s explana-
tions are little more than truisms — for example, that
“[i]ndividual people may have long lives and careers, and for-
eign governments and intelligence services may exist in per-
petuity” — without any indication that these particular PDBs
involve sources who could be threatened by disclosure at this
late date, or who could not be protected by redactions. If a
source were threatened, the CIA’s concerns would be quite
compelling, but we cannot discern that from such generaliza-
tions.

   [12] Nonetheless, Buroker does raise a related concern that
the revelation of sources that are even 40 years old could hin-
der the CIA’s current efforts to recruit individuals or govern-
ments as sources. Such potential sources may be frightened
off if they believe promises of confidentiality are subject to an
implicit time-based sunset clause at the discretion of the judi-
ciary. Courts have permitted the CIA to maintain the secrecy
of similarly dated material based on this concern. In Sims, for
example, the Supreme Court held that exemption 3 applied to
information that was about 20 to 30 years old. See Sims, 471
U.S. at 161. In so holding, the Court cited the necessity that
11354                        BERMAN v. CIA
the CIA be able to give potential sources “an assurance of
confidentiality that is as absolute as possible.” Id. at 175. The
Court explained that “[e]ven a small chance that some court
will order disclosure of a source’s identity could well impair
intelligence gathering and cause sources to ‘close up like a
clam.’ ” Id. The CIA therefore has a “compelling interest in
protecting both the secrecy of information important to our
national security and the appearance of confidentiality so
essential to the effective operation of our foreign intelligence
service.” Id. (citation and internal quotation marks omitted).
Following Sims’ instructions, other courts have permitted the
CIA to maintain the secrecy of fairly old documents. See
Maynard v. CIA, 

, 555 n.6 (1st Cir. 1993)
(“[T]he passage of thirty years, by itself, is insufficient to
require [the CIA] to disclose the information.”); Fitzgibbon,
911 F.2d at 763-74. The CIA’s assertion that disclosure of old
sources would detrimentally affect its ability to enlist new
sources is entitled to deference, see Sims, 471 U.S. at 179, and
warrants application of exemption 3 here.4

   [13] Accordingly, we hold that the requested PDBs are pro-
tected from disclosure by exemption 3. Critical to our analysis
— both with regard to the adequacy of the declarations and
the applicability of the exemption — is the uniqueness of the
PDBs as potentially the most important and classified intelli-
gence document in this nation’s intelligence apparatus. As
Buroker summarizes, “the PDB is the most highly selective
compendium of the most important intelligence available to
the U.S. Intelligence Community.” One observer has com-
mented more succinctly that the PDB is “the most restricted
  4
    This does not necessarily mean, as Berman suggests, that the CIA may
use this rationale to keep documents secret in perpetuity. Although our
broad deference to the CIA requires us to accept its assertion here, it is far
less plausible that current sources would “close up like a clam” if they
knew that their involvement with the CIA would be revealed four or five
generations later. At the very least, the CIA would need to provide a better
explanation of why documents of that vintage would need to remain
secret.
                        BERMAN v. CIA                      11355
document in Washington.” Bob Woodward, Bush at War 132
(2002). The extreme sensitivity of the PDB enhances the plau-
sibility of the CIA’s assertion that disclosure of the requested
PDBs could cause harm even 40 years after their generation.
It also justifies Buroker’s failure to describe in more detailed
terms the relationship between the requested PDBs and pro-
tected sources and methods. We do not suggest that PDBs are
categorically exempt from FOIA. But we are satisfied that
Buroker’s affidavit, when taken as a whole and viewed in
light of other evidence in the record, justifies the application
of exemption 3.

                               D.

   As an alternative to his assertion that disclosure of the con-
tents of the requested PDBs would reveal protected sources or
methods, Buroker submits that PDBs are themselves protected
intelligence methods. This is true, Buroker states, because
“[t]he PDB is part of the process by which the CIA advises
the President and his most senior advisors regarding the sub-
ject areas most important to them” and therefore intelligence
decisions “are directly affected by the PDB process.” Not-
withstanding the great deference we typically afford the
CIA’s affidavits, we reject Buroker’s attempt to create a per
se status exemption for PDBs.

   [14] Although PDBs will typically contain information that
reveals intelligence sources and methods, this does not mean
that PDBs themselves are intelligence methods. As can be
gleaned from the PDBs that are publicly available, PDBs are
nothing more than simple memoranda the CIA uses to com-
municate with the President. Historians have documented the
PDB process in such great detail that even if that process
could be deemed a “method,” that method has already been
fully disclosed to the public. If we were to accept the CIA’s
logic, then every written CIA communication — regardless of
content — would be a protected “intelligence method”
because it is a method that CIA uses in doing its work. The
11356                       BERMAN v. CIA
CIA would then be able to avoid entirely our requirement that
it provide a specific justification that explains why the partic-
ular document requested fits within exemption 3. See Hunt,
981 F.2d at 1119. We decline to adopt such a boundless defi-
nition, and instead hold that whether or not a particular docu-
ment used by the CIA in its ordinary course of business is an
intelligence method depends upon the content of the docu-
ment.
   [15] The CIA cites Aftergood v. CIA, 

,
562 (D.D.C. 2005), as supporting its expansive definition of
intelligence methods. In that case the district court held that
exemption 3 excused the CIA from disclosing its budget. The
court did not, however, hold that the budget itself is an intelli-
gence method, but rather that “intelligence budget information
relates to intelligence methods, namely the allocation, transfer
and funding of intelligence programs.” Id. (emphasis added)
(citation and internal quotation marks omitted). There, as
here, it was the content of the documents sought, not the doc-
uments themselves, that deserved protection under exemption
3.5
                                IV.
   We hold that exemption 3 excuses the CIA from release of
these requested PDBs. We reject, however, the CIA’s argu-
ment that the PDBs are themselves protected intelligence
methods.
   AFFIRMED.
  5
    Berman argues that if we reject the CIA’s claim that the PDBs are
themselves intelligence methods, we must also reject the CIA’s claim that
disclosure of the PDBs would reveal intelligence methods because the sec-
ond argument is inextricably intertwined with the first. Buroker’s declara-
tion specifically warns, however, that the requested PDBs “would disclose
specific intelligence methods, including technical collection methods.”
(Emphasis added.) Buroker’s prediction that intelligence methods would
be disclosed is not grounded in the premise that PDBs are themselves
methods.