dissenting with respect to Section IIB:
I agree with the majority’s analysis in all other respects, but cannot agree with its conclusion, in Section IIB, that Forest Guardians is barred from challenging the Agua-Caballos project on the ground that the agency changed its rationale when the project was challenged in court.
Under the Supreme Court’s seminal administrative law decision, SEC v. Chenery Corp., 318 U.S. 80, 87, 63 S.Ct. 454, 87 L.Ed. 626 (1943), courts may not affirm an administrative action on grounds different from those employed by the agency. In this case, the Forest Service approved the Agua-Caballos project under the 1982 rules, and supported the project throughout the administrative appeals process on the basis of those rules. It did the same when Forest Guardians challenged the decision in federal court and requested a preliminary injunction. In its brief opposing the preliminary injunction, however, the Forest Service for the first time suggested that the appropriate standard was in fact not the 1982 rules but the “best *1132available science” standard, which had been adopted in November 2000 on a transitional basis.1 Forest Guardians continued to argue that the 1982 rules governed, but also countered the Forest Service’s best available science argument by contending that, under Chenery, the Forest Service was foreclosed from now defending its decision on a standard it had not applied during its actual decision-making. P. Rep. Br. of Feb. 6, 2006 at 11-12. By the time Forest Guardians filed its next brief in the case, this court had held in other cases that the 1982 rules were not the appropriate standard; instead, best available science should govern. Forest Guardians brought this to the attention of the court, saying that while it still believed the 1982 rules were the appropriate standard, if the court determined that best available science applied, it would have to vacate the decision and remand to the agency, as the Forest Service had not professed to apply that standard below. P. Third Not. of Supp. Auth. at 4-5. The district court then convened a status conference during which it noted that best available science was the proper standard, that the Forest Service had not appeared to apply it, and that it was considering vacating and remanding. Instead of doing so, though, the court chose not to address the merits of the Chenery claim and vacate the decision but rather to find a failure to exhaust.
The majority holds that Forest Guardians may not object to the agency’s change of rationale because it did not raise this particular argument before the agency during the administrative appeal process. In my opinion, however, a Chenery claim is never barred for failure to exhaust administrative remedies. It is not logically possible for a Chenery claim to arise during the administrative process, because at that stage the shift in rationale has not yet occurred. Only when the agency offers a new rationale in district court or the district court affirms on the basis of a new rationale does the Chenery problem arise.2 *1133The question of what arguments Forest Guardians asserted before the agency below is therefore irrelevant to our inquiry into whether the district court upheld the agency’s action on an improper ground.
By faulting Forest Guardians for failing to raise the “best available science” standard during the administrative process, when the agency had not invoked that standard to support the proposed project, the majority misconceives the relation between exhaustion and Chenery claims. Exhaustion and Chenery are related in that each encourages arguments to be raised at the administrative level — exhaustion requires this of plaintiffs, and Chenery requires this of agencies. Here, it was the agency that failed to raise the “best available science” standard at the appropriate time, and yet we are removing the teeth from Chenery by using exhaustion to prevent the plaintiffs from calling them on it.
Because it is logically impossible for parties to raise a Chenery issue during the administrative process, the majority’s holding effectively gives a green light to Chenery violations, so long as agency lawyers are quick-witted enough to plead lack of exhaustion. Upon discovery that the agency’s rationale will not hold up in court, the lawyers offer a new one. When the plaintiff raises a Chenery objection, the agency lawyers cry, “no exhaustion.” It should work every time.
Deciding whether to excuse exhaustion is an “intensely practical” inquiry that should “be guided by the policies underlying the exhaustion requirement.” Bowen v. City of New York, 476 U.S. 467, 484, 106 S.Ct. 2022, 90 L.Ed.2d 462 (1986). Those policies include allowing an agency “an opportunity to correct its own errors, to afford the parties and the courts the benefit of its experience and expertise, and to compile a record which is adequate for judicial review.” Id. (quoting Weinberger v. Salfi, 422 U.S. 749, 765, 95 S.Ct. 2457, 45 L.Ed.2d 522 (1975)). Those concerns are not present here. There is no suggestion that Forest Guardians was withholding its Chenery claim so as to avoid agency review. The issue simply did not arise until the parties were before the district court. Nor do we need agency expertise when the issue is a straightforward legal one of whether an agency has changed the justification for its decision. Whether the forest project satisfies the “best available science” standard requires expertise; whether agency lawyers shifted their ground in district court does not.
The issue is not, as the majority seems to think, whether Forest Guardians could have anticipated that the 2000 transitional rules might be held to apply to the project. That may well be so. But the agency did not rely on the 2000 transitional rules when it approved the Agua-Caballos project. It relied on the 1982 regulations. Forest Guardians explained during the administrative process why it believed the agency action was unjustified under those 1982 regulations. It was not Forest Guardians’ obligation to put forward or to counter alternative rationales that might support the agency action. And once in court, Forest Guardians could expect that, under Chenery, the Forest Service would be limited to defending the project on the basis of the administrative rationale.
Our holding is even stranger when considered in light of some of our other prece*1134dents involving projects approved during the transitional period. Between 2000 and 2004, the Forest Service approved a number of forest projects, using the criteria of the 1982 rules. After this court held that the “best available science” standard of the 2000 transitional rules was applicable to these projects, the Forest Service began to defend these agency decisions under the new standard. See Ecology Center, Inc. v. U.S. Forest Service, 451 F.3d 1183, 1195 (10th Cir.2006) (holding that when the Forest Service made transition-period decisions under the 1982 rules rather than the best available science standard, those decisions were arbitrary and capricious). In Utah Environmental Congress v. Troyer, a divided panel of this court found the Forest Service’s approval of a project to be arbitrary and capricious under Chenery when the agency had made its decision under the 1982 rules, but the agency then successfully defended in federal district court under the best available science standard. 479 F.3d 1269, 1287-88 (10th Cir. 2007). The panel did so even though the environmentalist plaintiff had not even raised the Chenery claim in its appellate brief. See id. at 1289-90 (McConnell, J., dissenting) (arguing that the court should not reach the Chenery issue because plaintiff had raised it neither before the district court nor on appeal); see also Utah Environmental Congress v. Richmond, 483 F.3d 1127, 1136 (10th Cir.2007) (noting that in a conflict between the principle that we will not reverse a district court based on an argument not made by the plaintiff and the principle that we may not affirm an agency decision based on reasoning that the agency itself never considered, Troyer settled the issue in favor of the latter). The only difference between Troyer and the present case is that this time, after the Forest Service changed its rationale, the plaintiff raised the Chenery issue. Apparently that was its mistake. If Forest Guardians had never raised the claim, the Forest Service would not have had occasion to argue exhaustion, and the court would have been under an obligation to raise the Chenery issue sua sponte, as we did in Troyer and Richmond.
With all respect, I think we have gotten it backwards. Chenery claims are not subject to the requirement of exhaustion, but they are forfeited if not raised on appeal.
I agree with the majority’s resolution of Forest Guardians’ other claims, but I cannot join its holding in Section II.B regarding the exhaustion requirement. I would remand to the district court for consideration of the Chenery claim on the merits.
. There was a great deal of uncertainty about what rules applied during the transitional period. As late as 2004, the Forest Service was still applying the 1982 rules. Utah Environmental Congress v. Bosworth (UEC II), 439 F.3d 1184, 1189 (10th Cir.2006); see also Forest Watch v. United States Forest Service, 410 F.3d 115 (2d Cir.2005). For some time even our own court held that the 1982 rules governed management plans approved during the transitional period. Utah Environmental Congress v. Bosworth (UEC I), 372 F.3d 1219, 1221 n. 1 (10th Cir.2004); see also Utah Environmental Congress v. Bosworth (UEC II Superseded), 421 F.3d 1105, 1110-11 (10th Cir. 2005) , superseded by 439 F.3d 1184 (10th Cir. 2006) . The Forest Service applied the 1982 rules when approving the Agua-Caballos project and throughout the administrative appeals. In September 2004, however, the Department of Agriculture issued an interpretive rule stating that the "best available science" standard applied during the transitional period, not the 1982 rules. 69 Fed.Reg. 58,055-01 (Sept. 29, 2004). The Forest Service’s decision to move immediately to the transitional rules may have been triggered by this court's decision in UEC I that compliance with the detailed indicator species monitoring requirements of the 1982 rules must be achieved at the project level as a precondition to approval of any project. 372 F.3d at 1227. The prior interpretation seems to have been that these regulations required monitoring at the forest level, and provided that forestry management decisions be guided by the information this monitoring revealed about the health of the forest and the impact of various management practices. The shift in interpretation to the project level made many forest projects vulnerable to challenge under the 1982 regulations. If it intended to ease the path to judicial affirmance of forest projects by declaring, in 2004, that the "best available science" standard applied as of 2000, however, the Forest Service tripped over its own feet, because it made projects approved during the transitional period under the 1982 regulations vulnerable to challenge on the ground that they had not applied the new standards.
. The Forest Service now contends that a review of the administrative record will show that the agency in fact did apply the best *1133available evidence standard, though “not always explicitly labeling it as such.” Aple. Br. 36 n. 9. The district court expressed skepticism of this argument during its status conference, as the Forest Service "does not point in record to where it used best available science standard.” Clerk's Minutes of Jul. 19, 2006 Status Conf. at 2. This would be an issue for the district court to address while considering the merits of the Chenery claim; it is not a reason to bar the claim for a failure to exhaust.