ORDER AND JUDGMENT *

After Dr. Joseph Craine accused some of his colleagues of academic misconduct, Kansas State University investigated the matter, found the allegation unfounded, and effectively terminated Dr. Craine’s position. As all this unfolded, Dr. Craine submitted a complaint to the National Sci *872 ence Foundation (NSF), alleging that the . University was in the process of illegally retaliating against him, and seeking protection from the NSF under 41 U.S.C.

§ 4712, a statute that protects employees of federal grantees (like the University) from certain forms of reprisals, In a brief order, the NSF denied Dr. Craine’s claim under § 4712, stating:

Your complaint relates to communications you made to an academic journal regarding possible inaccuracies in another University employee’s manuscript. The Pilot Program expressly identifies the “persons or bodies” to whom a communication must be made to be considered a ..covered disclosure under the statute. 41 U.S.C. § 4712(a)(2)(A-G). Because your communications were not made to any of the persons or bodies enumerated in 41 U.S.C. § 4712(a)(2)(A-G) as statutorily required, the Foundation must deny your claim for relief. . Even if your communications had constituted a covered disclosure, we are persuaded that, based on the evidence before us, there was not a sufficient basis to conclude that the University subjected you to a reprisal prohibited by 41 U.S.C. § 4712(a).

It is this order we’re now asked to review.

Immediately though, we confront this problem: the agency hasn’t given us much to go on. When we engage in arbitrary and capricious review of agency action, we must “ascertain whether the agency examined the relevant data and articulated- a rational connection between the facts found and the decision made,” Olenhouse v. Commodity Credit Corp., 42 F.3d 1560, 1574 (10th Cir. 1994) (footnote omitted). That means “[t]he agency must make plain its coursé of inquiry, its analysis and its reasoning.” Id. at 1575. Yet the agency’s order in this case contains very little of that, offering much in the way of conclusion but little in the way of analysis or reasoning. To be sure, in their briefs the NSF and University offer a good deal of analysis and reasoning to support the NSF’s ultimate decision. And there might be merit to their arguments. But when it comes to. agency action, this kind of “[a]f-ter-the-fact rationalization by counsel in briefs or argument” just isn’t enough. Id. .The agency order itself must contain the agency’s analysis and reasoning, and our review is correspondingly limited to the reasons the agency offers for decision. Id.; see also Mickeviciute v. INS, 327 F.3d 1159, 1163 (10th Gir. 2003).

This matter is remanded to the NSF for further proceedings consistent with this opinion. The NSF’s motion for leave to file a surreply is denied.