RUTH BADER GINSBURG, Circuit Judge:

In this Freedom of Information Act (FOIA) case, plaintiff Critical Mass Energy Project (CMEP) seeks access to reports prepared by a utility industry consortium and voluntarily transmitted to the Nuclear Regulatory Commission (NRC or Commission). The district court granted the NRC's motion for summary judgment, holding that the reports in question could be withheld pursuant to 5 U.S.C. § 552(b)(4), FOIA's exemption for confidential, commercial information. Because the NRC has not established on the current record that exemption (b)(4) shelters the disputed reports, we vacate the judgment in the NRC's favor and remand the case for further proceedings.

I.

The Institute for Nuclear Power Operations (INPO) was formed in the aftermath of the Three Mile Island accident to promote, in the NRC's words, "improved safety and reliability in the operation of commercial nuclear power plants through a program of self-regulation by peer review." All 55 utility companies that operate or construct nuclear power plants in the United States are members of INPO, and finance its operations by their membership assessments.

As part of its "SEE-IN" (Significant Event Evaluation and Information Network) Program, INPO periodically produces three types of reports: Significant Event Reports (SERs), Significant Operating Experience Reports (SOERs), and Operation Maintenance Reminders (O MRs). SERs are produced whenever INPO determines, on the basis of information provided to it primarily by its member utilities, that a "significant" safety-related event has occurred at a nuclear power plant. The SER describes the event and its "possible causes and effects" in detail, and sets out "INPO's candid and critical analysis of the problems identified." SOERs contain more detailed follow-up analyses of these significant events, with more wide-ranging, generic recommendations concerning plant construction, design, and operation. O MRs concern operating problems deemed not significant enough to warrant the searching analysis included as part of the SERs and SOERs; O MRs "generally describe operating or maintenance difficulties encountered in specific nuclear plants and discuss the means used to solve the problems."

These reports, the focus of CMEP's FOIA request, are provided by INPO, at its expense, to the NRC under a 1982 Memorandum of Agreement. All INPO members and "participants," as well as the Nuclear Safety Analysis Center and the Nuclear Electric Insurance Limited, receive copies of the reports. Outside consultants and contractors may receive copies "where necessary in order to make appropriate use of the report, e.g., to take corrective action." Copies are also sent to vendors whose products are discussed in the reports. Apart from these routine distributions, the reports may be disclosed to third parties only upon INPO's specific consent.

The NRC resisted CMEP's request for access to these documents, claiming that they are covered by exemption 4 as "commercial or financial information obtained from a person and privileged or confidential." 5 U.S.C. § 552(b)(4). The district court agreed, and granted the NRC's motion for summary judgment. We vacate the judgment so ordered, concluding that the NRC, on the record thus far made, has not established that the documents sought are sheltered by the exemption claimed.

II.

To sustain its burden of demonstrating that these documents are within FOIA exemption 4, the NRC must show that the information contained therein is "(a) commercial or financial, (b) obtained from a person, and (c) privileged or confidential." Board of Trade v. Commodity Futures Trading Comm'n, 627 F.2d 392, 403 (D.C. Cir. 1980), quoting Getman v. NLRB, 450 F.2d 670, 673 (D.C. Cir. 1971). CMEP concedes that these reports were "obtained from a person," and the NRC admits that the information contained in the reports is neither "financial" nor "privileged"; therefore, we must decide, first, whether the information in the INPO reports is "commercial," and, if so, whether it is "confidential." We hold that the information is "commercial," but that the record, in its current state, does not permit a dispositive answer to the ultimate question: Is the information "confidential"?

A. The INPO Reports are "Commercial."

We agree with the district court's conclusion that, contrary to CMEP's contention, the reports contain "commercial" information: "INPO's constituent utility companies are assuredly commercial enterprises engaged in the production and sale of electrical power for profit [and] [t]he revelation of the details of the operations of their nuclear power plants . . . could materially affect their profitability in multiple ways." INPO itself is a not-for-profit enterprise, but INPO's non-profit status is not determinative of the character of the information it reports; information may qualify as "commercial" even if the provider's ( i.e., INPO's) interest in gathering, processing, and reporting the information is noncommercial. See Board of Trade, 627 F.2d at 405-06 (the "plain language" of exemption 4 "does not in any way suggest that this information must relate to the affairs of the provider"; the exemption is "sufficiently broad to encompass financial and commercial information concerning a third party").

In Public Citizen Health Research Group v. FDA, 704 F.2d 1280 (D.C. Cir. 1983), we held that manufacturers of intraocular lenses had a "commercial interest" in health and safety data submitted to the FDA because the data would be "instrumental in gaining marketing approval for their products." Id. at 1290. Comparably, the commercial fortunes of INPO's member utilities, and the vendors whose products are appraised in the INPO reports, could be materially affected by the disclosure of health and safety problems experienced during the operation of nuclear power facilities.

B. Are the INPO Reports "Confidential"?

Turning to the "confidential" nature of this information, the precedent that guides us establishes two prime requirements. First, the agency must demonstrate that the information it seeks to shield "would customarily not be released to the public by the person from whom it was obtained." Board of Trade, 627 F.2d at 404, quoting S.Rep. No. 813, 89th Cong., 1st Sess. 9 (1965). We conclude that the NRC has made this demonstration.

Second, the agency must demonstrate that "disclosure will harm a specific interest that Congress sought to protect by enacting the exemption." 9 to 5 Organization for Women Office Workers v. Board of Governors, 721 F.2d 1, 9 (1st Cir. 1983). In National Parks Conservation Ass'n v. Morton, 498 F.2d 765 (D.C. Cir. 1974), we focused upon two of those interests: the "need of government policymakers to have access to commercial and financial data," id. at 767; and the need to "protect persons who submit financial or commercial data to government agencies from the competitive disadvantages which would result from its publication." Id. at 768. We also noted that Congress had advanced these justifications for "the exemption of commercial material: (1) encouraging cooperation by those who are not obliged to provide information to the government and (2) protecting the rights of those who must." Id. at 769. Ultimately, we fashioned the following test:

[C]ommercial or financial matter is `confidential' . . . if disclosure of the information is likely . . . either . . . (1) to impair the Government's ability to obtain necessary information in the future; or (2) to cause substantial harm to the competitive position of the person from whom the information was obtained.

Id. at 770 (footnote omitted). We conclude that the NRC has not established the impairment of its future information-gathering ability prerequisite to affirming the entry of summary judgment in its favor.

We reserved in National Parks the question whether the government could sustain its exemption 4 burden by proving "impairment" of interests other than the two cited in the test we delineated. See id. at 770 n. 17 ("We express no opinion as to whether other governmental interests are embodied in this exemption."); see also Washington Post Co. v. HHS, 690 F.2d 252, 268 n. 51 (D.C. Cir. 1982) (similarly reserving judgment). We now hold, see infra Section II.B.3, that the NRC may invoke exemption 4 on the basis of interests other than the two identified in our National Parks test; here again, however, because we are uncertain of the precise position the NRC seeks to advance, we remand the matter for further proceedings.

1. Are the INPO Reports "Customarily . . . Released to the Public"?

We agree with the district court that the NRC has met its threshold burden on this issue. As noted earlier, see supra p. 280, the INPO reports receive fairly wide dissemination to individuals and organizations engaged in the nuclear power production process; nonetheless, receipt is premised upon compliance with INPO's stated policy of nondisclosure to third parties, and nothing in this record suggests that INPO's "limited distribution" policy is less than conscientiously observed. Because the class of recipients, while large, is well-defined, and because an explicit nondisclosure policy is apparently adhered to in practice, we hold that the INPO reports meet this first confidentiality threshold. Cf. Sharyland Water Supply Corp. v. Block, 755 F.2d 397, 399 (5th Cir. 1985) (audit reports available to water supply corporation's members held not confidential; "what five thousand people may obtain without even a pledge of nondisclosure is not confidential") (emphasis added).

2. Would Disclosure Impair the NRC's Ability to Obtain Necessary Information in the Future?

The NRC's "impairment" claim under this prong of the National Parks test, see supra p. 282, is not a model of clear statement. We endeavor next to examine its components.

First, the NRC points to "unequivocal statements" in the record "that INPO would not [voluntarily continue to] supply these reports to the NRC if the agency made them available to the public." CMEP, however, urges us to treat these "self-serving" representations on INPO's part with "considerable skepticism" inasmuch as "it runs directly counter to the interests of INPO and its members for INPO to jeopardize its relationship with the NRC by refusing to voluntarily release its safety reports."

Even if we were to assume that INPO would discontinue voluntary transmission to the NRC, however, it is not clear whether, or the extent to which, the Commission's future ability to obtain necessary information would be impaired. A "minor impairment," we have elsewhere observed, "cannot overcome the disclosure mandate of FOIA[;] the impairment [must be] significant enough to justify withholding the information." Washington Post, 690 F.2d at 269 (remanding case for the district court to make findings on "the extent" to which the government's ability to obtain information would be "impaired" by disclosure); accord Pacific Architects Engineers, Inc. v. Renegotiation Board, 505 F.2d 383, 385 (D.C. Cir. 1974) (same).

We note, in this regard, that INPO has not suggested that it would discontinue either production of the reports here at issue, dissemination of those reports to its members, or any other aspect of its SEE-IN program — let alone that INPO itself would disband — were we to reject definitively the NRC's FOIA exemption plea. Therefore, to show the requisite impairment of its information-gathering ability, the agency must persuade the court either (1) that cessation of INPO's voluntary submission of these reports would in fact deprive the agency of the information contained therein, or (2) that alternative available means for obtaining the INPO reports would entail a significant risk that the value of the submitted reports would decrease.

The district court made no findings on these questions, observing only that "it is preferable to have the INPO reports furnished to the NRC voluntarily, rather than delivered up under compulsion in circumstances less conducive to candor, accuracy, and timeliness." We agree, but were that proposition sufficient to overcome FOIA's disclosure mandate, our circuit's requirement that the agency provide a " detailed justification [of] the extent to which disclosure . . . will impair the government's ability to obtain necessary information [, supported by] specific factual or evidentiary material," Pacific Architects, 505 F.2d at 385 (emphasis added), would have little, or no, force. Why demand detail if the voluntary character of the submission suffices to shield the reports?

We note, first, the NRC's concession that it has ample statutory authority, under 42 U.S.C. § 2201(c), to compel the production of the INPO reports, either by "institut[ing] a program of periodic issuances of subpoenas" to INPO or its utility-members, or by requiring those utilities to submit the INPO reports as a condition of obtaining a license from the NRC. However, even if the NRC could compel production of all INPO reports, the question remains whether compelled production would entail qualitative impairment of the information contained therein. Our precedents recognize that "those who expect public dissemination of their remarks may well temper candor with a concern for appearances and for their own interests." Association for Women in Science v. Califano, 566 F.2d 339, 346 (D.C. Cir. 1977), quoting United States v. Nixon, 418 U.S. 683, 705, 94 S.Ct. 3090, 3106, 41 L.Ed.2d 1039 (1974); see also Washington Post, 690 F.2d at 268 ("If the government can enforce the disclosure obligation, and if the resultant disclosure is likely to be accurate, that may be sufficient to prevent any impairment.") (emphasis added). The NRC has asserted that disclosure would result in a "substantial risk that the reports . . . would become less valuable." But at this juncture, that assertion lacks clear focus and solid factual support.

The Commission's presentation leaves us uncertain as to the contours of this facet of the NRC's impairment claim. To explain our difficulty in fathoming the Commission's position, we describe in more detail the nature of the information contained in the INPO reports. INPO obtains most of the factual material on which its reports are based from information filed with the NRC by INPO's member utilities — information that the NRC routinely makes available to the public. INPO also obtains relevant data, however, from other, nonpublic sources, such as interviews with plant personnel during either routine site visits or detailed field investigations of particular safety-related events. In addition, the O MRs, see supra p. 280, may discuss "nonsafety related incidents that are not required to be reported [to] the NRC but that one of INPO's members has brought to [INPO's] attention, or . . . event[s] or occurrence[s] which [have] been reported to INPO by one of its vendor participants."

Quite apart from the factual material collected by INPO and included in its SEE-IN reports, INPO performs its own analyses of the reported events, e.g., trend analysis "to determine if patterns of recurrence among events indicate underlying problems requiring action," and sequential risk analysis "to examine the actual sequence and all possible sequences associated with an event in order to reveal problems with design and operational practices that might not be detected otherwise."

This dual aspect — factual and analytical — of the information the NRC receives from INPO suggests two ways in which the NRC's ability to obtain that information may be impaired by public disclosure. First, INPO's analyses may be conducted less vigorously, and in its brief to this court, the NRC so contends. The Commission asserts that "[i]f INPO knows that the reports will be publicly disclosed, its reports are less likely to be entirely candid and, therefore, helpful to the NRC." We find the record barren, however, of any support for this assertion; the record evidence to which the NRC's brief refers does not, in fact, even suggest that INPO would perform its analytical tasks with less candor or thoroughness post-disclosure than it does at present. Rather, that evidence speaks to a related, but distinct, claim, that the factual basis of the INPO reports will be impaired.

The claim that disclosure will impair the quality of the analysis is not implausible, particularly in light of the resemblance between INPO's analytical process and the agency "deliberative process" protected by exemption (b)(5), 5 U.S.C. § 552(b)(5). However, the agency's burden under FOIA requires more than plausibility. The NRC has failed to put forward the "detailed justification" our precedents require. Pacific Architects, 505 F.2d at 385; see also Orion Research Inc. v. EPA, 615 F.2d 551, 554 (1st Cir. 1980) (the "mere conclusory say-so of an agency that its ability to acquire information would be impaired will [not] suffice"; that conclusion must be "plausibly supported in some detail").

We mention in particular two shortcomings in the NRC's evidence. First, as noted above, see supra note 29, we find confusing and possibly contradictory evidence in the record concerning the presence vel non of any "base data" in the INPO reports that are otherwise not already in the agency's possession. Obviously, the NRC's ability to gather information it already independently receives cannot be "impaired" should the utilities become less than forthcoming in providing that information to INPO.

Second, even were we to credit the NRC's assertions that any lack of candor on the utilities' part would deprive it of some necessary information, we remind the agency of its burden to persuade the reviewing court that "the impairment is significant enough to justify withholding the information." Washington Post, 690 F.2d at 269. This requires the agency to specify, with greater precision than the NRC has here, the kinds of information it fears it will be deprived of in the future and the uses to which that information is currently put. Only then can a court intelligently gauge the nature and significance of the impairment that may accompany disclosure.

On remand, then, the NRC should be given the opportunity to document its necessarily uncertain prediction that disclosure would impair either the factual base, or analytical quality, of the INPO reports. To pass muster, the agency must spell out comprehensibly both the nature of the information of which it fears it will be deprived, and the nexus between the disclosure sought by CMEP and the anticipated impairment.

3. Will Disclosure Impair Other Interests Covered by Exemption 4?

As noted earlier, see p. 282 this circuit has not heretofore decided whether the test set forth in National Parks constitutes the exclusive means by which the government can sustain its exemption 4 burden, i.e., whether information can be justifiably withheld where disclosure would harm the agency in ways unrelated to its ability to gather information in the future. The First Circuit recently confronted this precise question and, after carefully examining both the relevant legislative history and case law, concluded that "[i]n view of the legitimate governmental interest of efficient operation, it would do violence to the statutory purpose of exemption 4 were the Government to be disadvantaged by disclosing information which serves a valuable purpose and is useful for the effective execution of its statutory responsibilities." 9 to 5 Organization, 721 F.2d at 11; see also id. at 12 (Breyer, J., dissenting) (agreeing with the court's expansion of the National Parks test while disagreeing with the application of the expanded standard to the case before the court).

We find the First Circuit's reasoning persuasive; however, we again conclude that the NRC has not sustained its burden of specifying the relevant "governmental interest which the Congress sought to protect by enacting exemption 4 [or demonstrating] how that interest will be harmed by public disclosure of the specific information which has been requested." Id. at 10.

The NRC's "impairment of agency efficiency" contention rests, it appears, on its assertion that "even if [it] could somehow get the same information it now receives from INPO in these reports" — e.g., by periodic issuances of subpoenas, see supra pp. 283-84 and notes 24-26 — "the sacrifice in the efficiency of its operations and the burdens it would necessarily bear in securing the information through other means would be too heavy." We note, however, that there is a very real possibility that the NRC's existing regulations encompass a requirement that its licensees submit these reports. The relevant prescription, 10 C.F.R. § 50.73(c), states:

The Commission may require the licensee to submit specific additional information beyond that required by [ 10 C.F.R. § 50.73(b)] if the Commission finds that supplemental material is necessary for complete understanding of an unusually complex or significant event.

The NRC's argument that it could not "demand that the licensees furnish it with copies of the reports as a condition to their status as licensees," is backed up only by an NRC staff member's statement that plausibly could be interpreted to mean either that the Commission simply has not made such a demand in the past, or that it lacks the authority to do so under the regulation. A linchpin of the NRC's argument in this case, we reiterate, is that the INPO reports are "necessary for complete understanding of . . . unusually complex or significant event[s]."

We therefore remand the question whether the Commission has authoritatively construed the regulation to be inapplicable to third-party reports. If the Commission or the lower court interprets the regulation as authorizing the NRC to obtain third-party reports, the court should then obtain the Commission's views on the application of the regulation to the facts of the case; that is, in view of the regulation's authorization to require additional material if necessary for the Commission's understanding of "an unusually complex or significant event," does the regulation cover SERs, SOERs, and O MRs? If so, the district court should afford the agency the opportunity to show, if it so chooses, that exercising its full authority under this regulation would damage some identifiable agency interest relating to program effectiveness or efficiency. If, on the other hand, the regulation does not authorize the Commission to obtain third-party reports, the district court should then consider the NRC's contention that resort to other means of obtaining this information would significantly impair "efficient operation[s]" or "effective execution of [NRC's] statutory responsibilities," 9 to 5 Organization, 721 F.2d at 11.

CONCLUSION

For the reasons stated, we vacate the district court's order directing summary judgment for the NRC. The Commission has not yet substantiated its claims that the disclosure it resists would impair either its ability to obtain necessary information in the future, or the efficient and effective performance of the regulatory responsibilities Congress has assigned to it. We therefore return this case to the district court for further proceedings consistent with this opinion.

It is so ordered.

. . . . .

(c) make such studies and investigations, obtain such information, and hold such meetings or hearings as the Commission may deem necessary or proper to assist it in exercising any authority provided in this chapter, or in the administration or enforcement of this chapter, or any regulations or orders issued thereunder. For such purposes the Commission is authorized to administer oaths and affirmations, and by subpoena to require any person to appear and testify, or to appear and produce documents, or both, at any designated place.