WILLIAMS, Circuit Judge:

Six years ago the Securities and Exchange Commission found that two accountants had engaged in "improper professional conduct" in violation of the Commission's Rule 2(e)(1)(ii), 17 CFR § 201.102(e)(1)(ii). After review in this court we remanded the case to the Commission, holding that it had failed to adequately explain its interpretation of the rule. Checkosky v. SEC, 23 F.3d 452, 454 (D.C. Cir. 1994) ("Checkosky I"). The Commission has evidently been unable to do so, voicing instead a multiplicity of inconsistent interpretations. In view of the Commission's inability to make any progress toward offering a single interpretation, and signs that the Commission is unlikely soon to make such progress, we are driven to the remedy reserved for rare cases of an agency's persistent failure to explain itself, and remand the case with instructions to dismiss the proceedings. See Greyhound Corp. v. ICC, 668 F.2d 1354 (D.C. Cir. 1981).

* * *

Because the facts are recounted at length in the separate opinions of Judges Silberman and Randolph in Checkosky I, we supply only a brief summary. In the first half of the 1980s petitioners David Checkosky and Norman Aldrich, accountants at Coopers Lybrand, performed a series of audits on behalf of Savin Corporation, a publicly traded company in the photocopier marketing business. During the years for which the audits were performed, Savin was trying (ultimately without success) to branch out into manufacturing by developing its own photocopier. Under generally accepted accounting principles ("GAAP"), costs of research and development must be expensed immediately rather than deferred. See Accounting For Research and Development Costs, Statement of Financial Accounting Standards No. 2, ¶ 12 (Fin. Accounting Standards Bd. 1974). But once RD is complete, a company may defer so-called "start-up" costs, see id. at ¶ 10, treating them as a capital item, presumably to be depreciated in due course. After consulting with Checkosky, Savin decided to defer the escalating costs of its design effort by categorizing them as start-up costs. The Commission later found that in financial statements filed with it for periods between May 1, 1980, and December 31, 1984, Savin improperly deferred $37 million in research and development costs in this fashion. In all cases Checkosky and Aldrich had reported that Savin's statements conformed with GAAP and that their own audits had been conducted according to generally accepted auditing standards ("GAAS").

The Commission initiated disciplinary proceedings against Checkosky and Aldrich in 1987, charging that the two accountants had engaged in "improper professional conduct" in violation of Rule 2(e)(1)(ii). An administrative law judge suspended them for five years from "practicing before the Commission," a broad term that encompasses preparation of any document for filing with the Commission. In 1992 the Commission affirmed the ALJ's finding that Checkosky and Aldrich had failed to observe GAAS and had improperly represented that Savin's financial statements complied with GAAP. In re David J. Checkosky Norman A. Aldrich, 50 S.E.C. 1180 (1992). The Commission stated that "a mental awareness greater than negligence is not required" to state a violation of Rule 2(e)(1)(ii), but "note[d]," as if in passing, that Checkosky and Aldrich's conduct "did in fact rise to the level of recklessness." Id. at 1197. The Commission thus affirmed the ALJ's finding that Checkosky and Aldrich violated Rule 2(e)(1)(ii). It reduced their suspension, however, from five years to two. Petitioners petitioned for review in this court on several grounds and we remanded to the Commission, holding that it had failed to adequately explain its interpretation and application of Rule 2(e)(1)(ii). Checkosky I, 23 F.3d at 454.

On January 21, 1997 the Commission issued an opinion on remand affirming the suspensions. In re David J. Checkosky Norman A. Aldrich, 7 Fed.Sec.L.Rep. (CCH) ¶ 74,386, at 63,421 (Jan. 21, 1997) ("1997 Op."). Checkosky and Aldrich again petitioned for review in this court, again claiming (among many other things) that the Commission had failed to articulate an intelligible standard for "improper professional conduct" under Rule 2(e)(1)(ii). Because we agree with this claim, we do not address the others.

* * *

In something of a tour de force, the Commission's 1997 opinion manages to both embrace and reject standards of (1) recklessness, (2) negligence and (3) strict liability — or so a careful (and intrepid) reader could find. It first appears to rely on a theory of recklessness. After a relatively brief survey of the facts, the opinion says: "We previously found that [Checkosky and Aldrich] engaged in improper professional conduct and that their conduct was reckless. We begin by explaining our reasons for this conclusion and why we continue to find their conduct reckless." 1997 Op. at 63,426. But after devoting several pages to an attempt to demonstrate petitioners' recklessness, the opinion abruptly forswears any reliance on that concept as an element of improper professional conduct under Rule 2(e)(1)(ii): "We believe that Rule 2(e)(1)(ii) does not mandate a particular mental state and that negligent actions by a professional may, under certain circumstances, constitute improper professional conduct." Id. at 63,430.

On review the Commission adhered to the second of these positions, disavowing any suggestion that recklessness is necessary for a violation of Rule 2(e)(1)(ii). At oral argument it likewise disclaimed reliance on recklessness as a standard for the substantive violation, see Transcript at 26-27, and hewed to the line adopted in its brief, treating recklessness as relevant only to the choice of sanction: "Only after concluding that petitioners had engaged in improper professional conduct did the Commission consider petitioners' mental state to determine whether to impose a sanction." Brief for Respondent at 49. Thus, although in fact the 1997 opinion began with a consideration of petitioners' mental state, the Commission's present position confirms that recklessness was not an element of its substantive charge.

With recklessness out of the picture, negligence would seem to be the most obvious remaining candidate. But the 1997 opinion failed to adopt an intelligible negligence standard. Instead, as we have already noted, it said only, "We believe that Rule 2(e)(1)(ii) does not mandate a particular mental state and that negligent actions by a professional may, under certain circumstances, constitute improper professional conduct." 1997 Op. at 63,430 (emphasis added). Elementary administrative law norms of fair notice and reasoned decisionmaking demand that the Commission define those circumstances with some degree of specificity. It has not done so.

The only further definition the Commission offered was its observation that negligent deviations from GAAS or GAAP will be held to violate Rule 2(e)(1)(ii) when they threaten the integrity of the Commission's processes. See, e.g., 1997 Op. at 63,429 ("Our conclusions about the propriety of particular professional conduct are driven by the impact on Commission processes of the specific facts presented in a given proceeding before us."). This is fine as an identification of one of the main underlying purposes of Rule 2(e), but not as a standard for determining violations of the rule in disciplinary proceedings. Accountants and attorneys practicing in the securities field will draw little comfort from the knowledge that their missteps will escape sanction as long as they do not "threaten the integrity of the Commission's processes." It is simply impossible to know in advance what sorts of negligent errors will meet this "standard"; we can imagine both narrow and potentially all-embracing constructions.

Finally, the Commission's opinion leaves open the possibility that a "standard" revolving around perceived danger to future processes might not even require a showing of negligence:

We wish to make clear, however, that the fact that GAAP and GAAS are professional standards against which we examine the conduct of accountants does not mean that every deviation from GAAP or GAAS is improper professional conduct warranting discipline under Rule 2(e)(1)(ii). Our processes are not necessarily threatened by innocent or even certain careless mistakes. At times, we have found improper professional conduct by accountants who engage in several deviations of [sic] GAAS or GAAP, or who deviated from GAAS or GAAP in more than one audit, or with more than one client. However, isolated failures may be so serious as to warrant discipline.

Id. at 63,432 (footnotes omitted). In the space of four short sentences this passage achieves impressive feats of ambiguity. The first sentence does not clearly rule out the possibility that non-negligent deviations from GAAS and GAAP could violate Rule 2(e)(1)(ii). The second suggests by negative implication that some innocent, i.e., non-negligent, mistakes will be held to transgress the Rule. And the third and fourth explicitly reserve authority to penalize even an "isolated" deviation from GAAS or GAAP if "serious" enough, though as we have noted the relevant characteristics of seriousness are nowhere defined.

Not only does the opinion on remand provide no clear mental state standard to govern Rule 2(e)(1)(ii), it seems at times almost deliberately obscurantist on the question. For example, it observes unhelpfully that "[i]mproper professional conduct by accountants encompasses a range of conduct." Id. at 63,429. Later the Commission cites several cases in which it has imposed sanctions for improper professional conduct, concluding with the following summary: "While the acts in each case demonstrated varying degrees of care or mental state, we concluded in each that the accountant had improperly certified that financial statements complied with the applicable auditing requirements and that the resulting financial statements could not be relied upon." Id. at 63,430. However legitimate and, indeed, essential the Commission's concern about unreliable financial statements may be, it is no substitute for a clearly delineated standard. Instead, the Commission's statements come close to a self-proclaimed license to charge and prove improper professional conduct whenever it pleases, constrained only by its own discretion (combined, perhaps, with the standards of GAAS and GAAP).

In summary, the Commission's opinion yields no clear and coherent standard for violations of Rule 2(e)(1)(ii). Although we owe "substantial deference to an agency's interpretation of its own regulations," Thomas Jefferson Univ. v. Shalala, 512 U.S. 504, 512 (1994), we cannot defer to an agency when "we are at a loss to know what kind of standard it is applying or how it is applying that standard to this record." United Food and Commercial Workers International Union v. NLRB, 880 F.2d 1422, 1435-36 (D.C. Cir. 1989).

Of course the agency was in a bind. On the one hand, reliance on negligence had its perils. Judge Silberman had noted in Checkosky I that adoption of a negligence standard might be ultra vires; given that much of the substantive law enforced by the Commission requires a showing of scienter, use of a negligence standard to penalize professionals might be viewed as a back-door expansion of its regulatory oversight powers. See Checkosky I, 23 F.3d at 459 (Silberman, J.). On the other hand a recklessness standard brought its own risks: there might not be substantial evidence to support a finding of reckless conduct. Nevertheless the Commission had to make a choice. There is no justification for the government depriving citizens of the opportunity to practice their profession without revealing the standard they have been found to violate. And, indeed, there is no reason for us to address the issues (whether of a legal or factual character) raised by each of the hypothetical stances that the Commission might have adopted.

When an agency utterly fails to provide a standard for its decision, it runs afoul of more than one provision of the Administrative Procedure Act. As Judge Silberman noted in the first appearance of this case before us, we have held on occasion that an "agency's failure to state its reasoning or to adopt an intelligible decisional standard is so glaring that we can declare with confidence that the agency action was arbitrary and capricious." Checkosky I, 23 F.3d at 463 (Silberman, J.). In addition, an agency violates the APA when it fails to include in its adjudicatory decision a meaningful "statement of findings and conclusions, and the reasons or basis therefor, on all the material issues of fact, law, or discretion presented on the record." 5 U.S.C. § 557(c)(3)(A). On at least these criteria, the Commission has defaulted.

There remains the question of remedy. In Greyhound Corp. v. ICC, 668 F.2d 1354 (D.C. Cir. 1981), we found for the second time that the Interstate Commerce Commission had failed to adequately justify its decision to maintain securities jurisdiction over the Greyhound holding company in light of ICC precedents that seemed to preclude such a ruling. In the second go-round, we noted that eight years had passed since Greyhound first requested relief from the agency. "The Commission," we concluded, "has had ample time and opportunity to provide a reasoned explanation of the decision to continue the exercise of its securities jurisdiction. We find no useful purpose to be served by allowing the Commission another shot at the target." Id. at 1364.

It is true, as then-Judge Thomas pointed out in a concurring opinion in 1991, that application of the Greyhound remedy ought to be "reserved for truly extraordinary situations," since "legitimate concerns about judicial overreaching always militate in favor of affording the agency just one more chance to explain its decision." Tennessee Gas Pipeline Co. v. FERC, 926 F.2d 1206, 1214 (D.C. Cir. 1991) (Thomas, J., concurring). Our use of the device has indeed been maximally sparing — Judge Thomas could not find a case in which we had applied it in the decade between Greyhound and Tennessee Gas Pipeline, and we have not found an example from the seven years since that decision. The case has, however, been followed in the Third Circuit. In Marshall v. Lansing, 839 F.2d 933 (3d Cir. 1988), that court upheld a district court order requiring the Parole Commission to recategorize a prisoner's offense as involving less than a kilogram of cocaine, after the Commission on remand had failed to offer an acceptable basis for its contrary finding: "When a court has already remanded a case to an administrative agency for failure to explain adequately its decision, and the agency, on remand, again fails to provide a reasoned basis for its conclusions, a reviewing court can set aside the agency's decision . . . ." Id. at 945 (citing Greyhound).

This case presents the sort of extraordinary situation for which application of Greyhound is reserved. The disciplinary proceeding has dragged on for more than ten years. It is based on events that occurred as long ago as 1980. Our 1994 decision set the Commission a straightforward task: to "choose its standard and forthrightly apply it to this case." Checkosky I, 23 F.3d at 462 (Silberman, J.). It has signally failed to do so.

Moreover, there are strong signs that the Commission is unlikely to settle on a uniform theory as to the necessary mental state for a violation of Rule 2(e)(1)(ii) anytime soon. In In re Robert D. Potts, 7 Fed.Sec.L.Rep. (CCH) ¶ 74,479, at 63,597 (Sept. 24, 1997), two Commissioners found that a concurring partner in an accounting firm had violated Rule 2(e)(1)(ii) by failing to comply with GAAS and representing that his client's financial statements accorded with GAAP.

Id. at 63,604 n. 40. Because these two Commissioners found the accountant's conduct reckless, they declined to address whether mere negligence can constitute improper professional conduct under Rule 2(e)(1)(ii). Id. at 63,605 n. 44. In his concurrence, Commissioner Johnson — who dissented from the 1997 Checkosky opinion — agreed that Potts had acted recklessly but expressly stated his view that scienter was required to establish a violation of Rule 2(e)(1)(ii). Id. at 63,608 n. 1. In dissent, Commissioner Wallman — who did not participate in Checkosky — expressed the view that a charge of improper professional conduct cannot be based on "mere negligence." Id. at 63,609-13. Of course, a decision such as Potts, issued after the one under review, cannot be cited in support of a claim that the one being reviewed is inconsistent with agency precedent. See, e.g., MacLeod v. ICC, 54 F.3d 888, 892 (D.C. Cir. 1995). But we can take judicial notice of it to gauge the futility of allowing the current proceedings to drag on into another round in the hope that the Commission will do what it should have done in each of the earlier rounds. Cf. Oil, Chemical and Atomic Workers Int'l Union v. NLRB, 46 F.3d 82, 92-93 (D.C. Cir. 1995) (refusing to sustain agency action where there is no sustainable view, or set of views, supported by a majority of the agency). In view of the Commission's repeated failure to articulate a discernible standard for violations of Rule 2(e)(1)(ii), the extraordinary duration of these proceedings, and the apparent unlikelihood of a clear resolution on remand, we conclude that it would be futile to allow the SEC a third "shot at the target." Greyhound, 668 F.2d at 1364.

The case is remanded with instructions to dismiss the charge against petitioners.

So ordered.