Opinion for the Court filed by Circuit Judge HARRY T. EDWARDS.

*1120 HARRY T. EDWARDS, Circuit Judge:

At issue in this case is a three-page document with accompanying cover letter (the “Review Letter”) containing the comments of two referees for the American Journal of Epidemiology (“Journal”). The Review Letter reviewed a report (“Report”) that a staff member of the Centers for Disease Control (“CDC”) had submitted for possible publication in the Journal. The Formaldehyde Institute (“Institute”) requested a copy of the Review Letter from the Department of Health and Human Services (“HHS”). HHS, acting on behalf of its constituent member CDC, refused, invoking “Exemption 5” of the Freedom of Information Act (“FOIA”), which protects certain “inter-agency or intra-agency memorandums or letters” from disclosure. See 5 U.S.C. § 552(b)(5) (1988). The Institute brought an action in the District Court to secure release of the Review Letter under FOIA; on cross-motions for summary judgment, the trial court granted the Institute’s request to compel HHS to release the document. See Formaldehyde Inst. v. HHS, Civ. Action No. 87-3266 (D.D.C. Sept. 6, 1988), reprinted in Joint Appendix (“J.A.”) 44 (Order). HHS now appeals to this court.

The law speaks clearly on this issue. An agency may withhold a document under Exemption 5 when it is both predecisional and deliberative. The Review Letter is predecisional because it preceded the agency’s decision whether and in what form to publish the Report. The letter is part of HHS’ deliberative process because the agency secured review commentary in order to make that decision. Releasing materials that satisfy these Exemption 5 criteria could seriously hamper the efforts of CDC to fulfill its clear Congressional mandate to conduct and publish scientific research for the public benefit. Because the District Court erred in granting summary judgment for the Institute, we reverse. The case will be remanded to the trial court for entry of summary judgment in favor of HHS.

I. Baokground

In the course of his official duties as a staff researcher at HHS, Leslie Stayner submitted the Report to the Journal to be considered for publication. The Report, “A Retrospective Cohort Mortality Study of Workers Exposed to Formaldehyde in the Garment Industry,” contains information about the harmful effects of formaldehyde on certain classes of workers. Pursuant to its normal review process, the Journal sent the Report to outside referees. In light of the reviews that were received, the Journal decided not to publish Stayner’s report. The reviews constituting the Review Letter were then sent to Stayner and CDC along with the decision not to publish.

On June 1, 1987, the Institute, invoking FOIA, requested copies of all agency records of contact between Stayner, members of the National Institute for Occupational Safety and Health (“NIOSH”), another of the constituent organizations of HHS (the third of which is the Public Health Service (“PHS”)), and the Journal, related to “publication or rejection” of the Report. See J.A. 21 (Institute letter). On July 2, 1987, HHS wrote back to the Institute stating that it was withholding the information based on Exemption 5. HHS subsequently denied the Institute’s administrative appeal of July 14, 1987, stating that the Journal was “functionally equivalent to agency staff and thus the Journal’s recommendations to agency decisionmakers should be protected.” Again the agency relied principally on the authority of Exemption 5 in withholding disclosure of the Review Letter. See J.A. 25 (HHS letter).

The Institute seeks the Review Letter in order to use it to challenge the findings of the Report, which, subsequent to the Journal’s rejection, was published in some form in another journal. See Brief for Appellee at 10 n. 5 (citing report published in 13 American Journal of Industrial Medicine 667 (1988)); id. at 18 (noting that “[t]he Institute has submitted extensive criticisms of the [Report]” to several government agencies). The Journal has a policy, designed to protect the integrity of the review process, of declining to release such letters to the public. See Comstock Declaration ¶ 5, reprinted in J.A. 37.

*1121 In an order and memorandum opinion filed on September 6, 1988, the District Court granted summary judgment in favor of the Institute. See Formaldehyde, slip op. at 1-6, reprinted in J.A. 39-44. The trial judge acknowledged that the Institute’s claim presented a “very close case,” in part because HHS had “shown that the receipt of comments [from outside referees] is an expected result of the submission of an article for publication.” Id. at 3, reprinted in J.A. 41 (citing HHS declar-ants). Nevertheless, the trial judge reasoned that Exemption 5 did not protect the Review Letter because the Journal was neither part of HHS nor an outside consultant:

[T]he Department did not retain or commission the [Journal] or the reviewers it selected as an outside consultant to give advice on matters of policy. Instead, the [Journal] acted as an independent publisher. The [Journal] selected individuals to review an article submitted for publication, determined that the article should not be published in that journal, and returned the reviewers’ comments to the author. Critically, the comments were generated as part of the [Journal’s] internal process of review, and not as part of the Department’s consultative process. The [Journal] solicited the comments to help it decide whether or not to publish the article. The fact that the [Journal] intended those comments to be confidential did not transform the [Journal] into the functional equivalent of an arm of the Department.

Id. at 3-4, reprinted in J.A. 41-42. In compelling disclosure of the Review Letter, however, the trial court allowed HHS to withhold the names of the reviewers in order to guard against any “possible chilling effect” on the review process. See id. at 4, reprinted in J.A. 42. 1 This appeal followed.

II. Analysis

A. Applicable Exemption 5 Criteria

Under Exemption 5, an agency 2 may withhold from disclosure “inter-agency or intra-agency memorandums or letters which would not be available by law to a party other than an agency in litigation with the agency.” 5 U.S.C. § 552(b)(5) (1988). Courts have “construed this exemption to encompass the protections traditionally afforded certain documents pursuant to evidentiary privileges in the civil discovery context,” including “ ‘materials which would be protected under the attorney-client privilege, the attorney work-product privilege, or the executive “deliberative process privilege.” ’ ” Taxation With Representation v. IRS, 646 F.2d 666, 676 (D.C.Cir.1981) (quoting Coastal States Gas Corp. v. Department of Energy, 617 F.2d 854, 862 (D.C.Cir.1980)) (citations omitted); see also National Labor Relations Bd. v. Sears, Roebuck & Co., 421 U.S. 132, 149, 95 S.Ct. 1504, 1515, 44 L.Ed.2d 29 (1975) (Exemption 5 protects documents that private party could not obtain in civil discovery).

In order to qualify for Exemption 5 protection, an agency’s materials must be both “predecisional” and a part of the “deliberative process.” See Sears, Roebuck, 421 U.S. at 151-52, 95 S.Ct. at 1516-17 (discussing “predecision” criterion) (citing cases); Renegotiation Bd. v. Grumman Aircraft, 421 U.S. 168, 184, 95 S.Ct. 1491, 1500, 44 L.Ed.2d 57 (1975) (same); Coastal States Gas Corp. v. Department of Energy, 617 F.2d 854, 866 (D.C.Cir.1980) (privilege “variously described as predecisional or deliberative”); Wolfe v. HHS, 839 F.2d 768, 774 (D.C.Cir.1988) (privilege limited to “materials which are both predecisional and deliberative”). A document may be “predecisional” and still fail to fall within the confines of Exemption 5 if it is not part of the “deliberative process.” In other words, while these two criteria are not *1122 mutually exclusive, neither are they coterminous in their reach.

A “predecisional ” document is one “prepared in order to assist an agency decisionmaker in arriving at his decision,” Grumman Aircraft, 421 U.S. at 184, 95 S.Ct. at 1500, and may include “recommendations, draft documents, proposals, suggestions, and other subjective documents which reflect the personal opinions of the writer rather than the policy of the agency,” Coastal States, 617 F.2d at 866. A predecisional document is a part of the “deliberative process,” if “the disclosure of [the] materials would expose an agency's decisionmaking process in such a way as to discourage candid discussion within the agency and thereby undermine the agency’s ability to perform its functions.” Dudman Communications Corp. v. Department of the Air Force, 815 F.2d 1565, 1568 (D.C.Cir.1987).

Several of our cases have described the scope of Exemption 5. In Ryan v. Department of Justice, 617 F.2d 781 (D.C.Cir.1980), for example, we held that congressional responses to Justice Department questionnaires are protected from disclosure under Exemption 5, even though members of Congress are not within the compass of the term “agency” under FOIA.

“The exemption,” we explained,

was created to protect the deliberative process of the government, by ensuring that persons in an advisory role would be able to express their opinions freely to agency decision-makers without fear of publicity. In the course of its day-to-day activities, an agency often needs to rely on the opinions and recommendations of temporary consultants, as well as its own employees. Such consultations are an integral part of its deliberative process; to conduct this process in public view would inhibit frank discussion of policy matters and likely impair the quality of decisions.

Id. at 789-90 (citations omitted). We have recognized that FOIA exemptions are to be construed “as narrow[ly] as [is] ‘consistent with efficient Government operation.' ” Id. at 790 (quoting S.Rep.No. 813, 89th Cong., 1st Sess. 9 (1965)); accord FBI v. Abramson, 456 U.S. 615, 621, 102 S.Ct. 2054, 2059, 72 L.Ed.2d 376 (1982). Nonetheless, as Ryan says, it is “unquestionably” true that

efficient government operation requires open discussions among all government policy-makers and advisors, whether those giving advice are officially part of the agency or are solicited to give advice only for specific projects. Congress apparently did not intend “inter-agency” and “intra-agency” to be rigidly exclusive terms, but rather to include any agency document that is part of the deliberative process.

Ryan, 617 F.2d at 790.

In CNA Financial Corp. v. Donovan, 830 F.2d 1132 (D.C.Cir.1987), cert. denied, 485 U.S. 977, 108 S.Ct. 1270, 99 L.Ed.2d 481 (1988), we noted and followed the approach enunciated in Ryan. As we pointed out:

courts have repeatedly found that a privilege attaches to reports of outsiders commissioned by an agency to perform agency work, when such reports would be protected if compiled within the agency itself. Whether the author is a regular agency employee or a temporary consultant is irrelevant; the pertinent element is the role, if any, that the document plays in the process of agency deliberations. If information communicated is deliberative in character it is privileged from disclosure, notwithstanding its creation by an outsider.

Id. at 1161-62 (citations omitted). Moreover, we stated, “federal agencies occasionally will encounter problems outside their ken, and it clearly is preferable that they enlist the help of outside experts skilled at unravelling their knotty complexities.” Id. at 1162 (citation omitted).

Similarly, in Russell v. Department of the Air Force, 682 F.2d 1045 (D.C.Cir.1982), we allowed the Air Force to use Exemption 5 to withhold portions of a manuscript of a history that it later published without those portions. We stated — quite instructively for the instant case — that “[t]he policies embodied in [Exemption 5] are as applicable to the [agency’s] editorial *1123 review process as they are to other agency deliberations that precede agency decisions.” Id. at 1049.

Finally, in Dudman we held that Exemption 5 prevented a radio broadcaster from using FOIA section 552, 5 U.S.C. § 552 (1988), to obtain a draft manuscript history of the Air Force role in Vietnam. Reaffirming Russell’s analysis, we concluded in Dudman that “disclosure of editorial judgments — for example, decisions to insert or delete material or to change a draft’s focus or emphasis — would stifle the creative thinking and candid exchange of ideas necessary to produce good historical work.” Id. at 1569. Summing up a discussion of much of this court’s Exemption 5 teachings, we observed that

release of materials plausibly labelled “deliberative” will occasionally reveal nothing about an agency deliberative process.... Conversely, the release of materials plausibly labelled “factual” will occasionally reveal much about that pro-cess_ Courts therefore began to focus less on the nature of the materials and more on the effect of the materials’ release: the key question in Exemption 5 eases became whether the disclosure of materials would expose an agency’s deci-sionmaking process in such a way as to discourage candid discussion within the agency and thereby undermine the agency’s ability to perform its functions.

815 F.2d at 1568.

B. Application of Exemption 5 Criteria

The Institute argues, and the District Court concluded, that because the Journal is neither part of HHS nor one of HHS’ consultants, the Review Letter is not entitled to Exemption 5 protections. See Brief for Appellee at 6-11; Formaldehyde, slip op. at 3-4, reprinted in J.A. 41-42. In support of this argument the Institute asserts that “the contours of ‘agency’ for the purposes of Exemption Five” are limited to cases in which an agency retains a temporary consultant or in which other government entities report to the agency. See Brief for Appellee at 8. It relies on Ryan and Donovan for the proposition that Exemption 5 is limited to cases where “the agency solicits the materials.” See id. The Institute also asserts, again relying on Ryan, that the documents at issue here did not arise from the agency deliberative process, see id. at 9, that they were “neither predecisional nor deliberative in nature,” see id. at 11-14, and that HHS was unable to state any “policy decision that is the subject of deliberation,” see id. at 12. We now address these arguments in light of the cases discussed above.

First, there is no doubt that the Review Letter is “predecisional,” in that it is a recommendation (with suggestions) regarding an article’s suitability for publication. The Review Letter aids CDC authors who must decide whether and to what extent to edit an article, and it assists agency decisionmakers who are authorized to determine whether and where to publish an article. See Roberts Declaration 119, 11, reprinted in J.A. 18-19; Tabor Declaration ¶ 5, reprinted in J.A. 35. The Review Letter thus easily qualifies as a predecisional document under Grumman, see 421 U.S. at 184, 95 S.Ct. at 1500, and Coastal States, see 617 F.2d at 866.

As to the question regarding the nature of the relationship between the outside reviewers and CDC, Ryan established and Donovan recently iterated that “inter-agency” and “intra-agency” are not “rigidly exclusive terms,” but rather embrace “any agency document that is part of the deliberative process.” See Ryan, 617 F.2d at 790; Donovan, 830 F.2d at 1161-62. Thus, both the Institute and the District Court err in focusing on the absence of any formal relationship between the Journal’s reviewers and HHS. As noted above, “[wjhether the author is a regular agency employee or a temporary consultant is irrelevant; the pertinent element is the role, if any, that the document plays in the process of agency deliberations.” Donovan, 830 F.2d at 1161. If the Review Letter is “deliberative in character,” then it may come within the confines of Exemption 5 “notwithstanding its creation by an outsider.” See id. at 1161-62. The pertinent issue here is what harm, if any, the Review *1124 Letter’s release would do to HHS’ deliberative process.

“[T]he first step in determining whether disclosure would harm the deliberative process is to examine the context in which the materials are used.” Wolfe, 839 F.2d at 774. Accordingly, we must discern what role publication in scientific journals plays in HHS’ deliberative process. On this point, Congress has instructed the Secretary of HHS to

conduct ... and encourage, cooperate with, and render assistance to other appropriate public authorities, scientific institutions, and scientists in the conduct of, and promote the coordination of, research, investigations, experiments, demonstrations, and studies relating to the causes, diagnosis, treatment, control, and prevention of physical and mental diseases and impairments of man.

42 U.S.C. § 241(a) (Supp. V 1988). Further, Congress has also authorized the Secretary to “collect and make available through publications and other appropriate means, information as to, and the practical application of, such research and other activities.” 42 U.S.C. § 241(a)(1) (1982).

Thus, Congress has directed HHS to make precisely the kind of “deliberative” decision HHS made as a result of a process that involved HHS’ receipt of the Review Letter. Indeed, a deliberative decision by HHS (through its constituent CDC) was central to the process of journal submission and review at issue here: namely, the deliberative decision about whether and in what form to publish the Report in the name of the agency.

The Institute acknowledges that, under Ryan and Donovan, the materials an agency solicits are “paradigmatically” entitled to Exemption 5 coverage. See Brief for Appellee at 8. The Institute argues, however, that in this case there can be no Exemption 5 coverage because the outside reviews were not “solicited” by HHS. We reject this position as too short-sighted in its view of Exemption 5. For one thing, it is an undisputed fact that “receipt of comments [from outside reviewers] is an expected result in the submission of an article for publication.” See Formaldehyde, slip op. at 3, reprinted in J.A. 41; see also Tabor Declaration ¶ 5, reprinted in J.A. 35. Furthermore, it is undisputed that reviewers’ comments are “expected to be confidential.” Formaldehyde, slip op. at 3, reprinted in J.A. 41. The agency does not “solicit” reviews in the sense that it contracts to receive them, but it does actively seek to do business with journals from which reviews are both expected and then used by CDC to determine whether and in what form to publish articles in the name of the agency. This arrangement reflects a mutual understanding between the agency and journals that provide confidential reviews regarding how the agency will use the reviews. The existence of such an arrangement is more than enough to hold that the Review Letter is a part of the deliberative process of the agency.

Finally, on the record in this case, it is indisputable that disclosure of reviewers’ comments would seriously harm the deliberative process. For example, HHS submitted the affidavit of Dr. Joseph E. Rail, who is Deputy Director for Intramural Research of the National Institutes of Health, to attest to the harms of disclosure. Dr. Rail is a research scientist who throughout his distinguished career has published many articles in scientific journals, administered research for the federal government, served as editor for two scientific journals and served as a reviewer for others. See Rail Declaration ¶¶ 1-2, reprinted in J.A. 27 (with attachment). His affidavit provided a detailed account of the reasons for which scientific journals have a review process in which both reviewers’ names and comments are shared only with the journal and the prospective author. See id. ¶¶ 4-5, reprinted in J.A. 28-29. He stated that “the comments of the reviewers are considered to be advice, constructive criticism and guidance with respect to revision of the manuscript so as to make the manuscript worthy of publication. This advice is generally relied upon by the author(s).” Id. 11 6, reprinted in J.A. 29. Additionally, he asserted that

*1125 the normal workings of review and publication of scientific articles would, in my opinion, be seriously compromised if comments of manuscript reviewers were to be released to the public. This would cause harm to government agencies in at least two ways: Government employees who must publish as part of their job responsibilities would no longer receive the candid, constructive advice that contributes to the author’s efforts to produce the best product possible. The journals, which have a mutual relationship with government scientists, would suffer because they could not be assured that the articles submitted by government scientists were the best possible product. In my opinion, release of reviewers’ comments would have a chilling effect on any comments submitted. Indeed, if reviewers thought that their comments would be available to competitors or opponents of the author(s), they might slant or soften their statements if they believed the article to be generally good.

Id. 117, reprinted in J.A. 30. 3

This and other like declarations offered by HHS were not contested by the Institute. Thus, the undisputed factual record clearly establishes that HHS personnel acting in light of the agency’s Congressional mandate must regularly rely on the comments of expert scientists to help them evaluate the readiness of agency work for publication. In that sense they must “rely on the opinions and recommendations of temporary consultants.” Ryan, 617 F.2d at 789. There is also unrefuted evidence in the record that release of reviewers’ editorial comments would very likely have a chilling effect on either the candor of potential reviewers of government-submitted articles or on the ability of the government to have its work considered for review at all. Furthermore, a government author is likely to be less willing to submit her work to a refereed journal at all if critical reviews could come to light somewhere down the line. This would result in the publication of inferior work in (presumably) inferi- or and less widely circulated journals, regardless of whether the identity of the critic was made known. As the Supreme Court has said with respect to Exemption 5, “ ‘[hjuman experience teaches that those who expect public dissemination of their remarks may well temper candor with a concern for appearances ... to the detriment of the decisionmaking process.’ ” Sears, Roebuck, 421 U.S. at 150-51, 95 S.Ct. at 1516-17 (quoting United States v. Nixon, 418 U.S. 683, 705, 94 S.Ct. 3090, 3106, 41 L.Ed.2d 1039 (1974)) (emphasis in original) (emendations in original). “Manifestly,” the Court continued, “the ultimate purpose of this long-recognized privilege is to prevent injury to the quality of agency decisions.” Id. 421 U.S. at 151, 95 S.Ct. at 1516.

III.- ConClusion

On the record of this case, we find that the Review Letter was both “predecisional” and a part of the agency’s “deliberative process.” Therefore, we hold that, pursuant to Exemption 5, HHS may withhold the Review Letter from disclosure under FOIA. Accordingly, the case is hereby remanded to the District Court for entry of summary judgment in favor of HHS.

. Since the Journal, in accord with its policy, had not forwarded the reviewers’ names along with the Review Letter, the HHS did not have and could not have released that information in any event.

. The CDC, like its HHS co-constituents PHS and NIOSH, is an "agency" subject to the application of FOIA. See 5 U.S.C. § 551 (1988).

. The Institute argues that releasing the Review Letter while maintaining the reviewer’s anonymity will “limit or preclude any hypothetical chilling effect." See Brief for Appellee at 14. However, the last sentence of the Rail affidavit quoted above makes clear that even if their names were withheld, reviewers who were favorably inclined to an article are likely to give less than candid comments, which will hurt the quality of agency publications. Moreover, although the Court in Sears, Roebuck, did not directly address the question of release of comments with or without names, the Court did note that “ 'those who expect public dissemination of their remarks,’" not only of their remarks along with their identities, “‘may well temper candor with a concern for appearances...'" Sears, Roebuck, 421 U.S. at 150, 95 S.Ct. at 1516 (quoting United States v. Nixon, 418 U.S. 683, 705, 94 S.Ct. 3090, 3106, 41 L.Ed.2d 1039 (1974)) (emphasis added).