OPINION
WIESE, Judge.The plaintiff in this suit, McLean Hospital Corporation (the “hospital”), seeks reimbursement of costs incurred in providing inpatient mental health care to a beneficiary of the Civilian Health and Medical Program of the Uniformed Services (“CHAMPUS”). Under the terms of the legislation defining the CHAMPUS program, 10 U.S.C. §§ 1071-1103 (1982 & Supp. V 1987), coverage of inpatient health care costs is limited to 60 days per patient per year unless a waiver of this restriction has been authorized by the Secretary of Defense because of “extraordinary medical or psychological circumstances that are confirmed by review by a non-Federal health professional pursuant to regulations prescribed by the Secretary of Defense.” 10 U.S.C. § 1079(i)(4) (Supp. V 1987).
An application for waiver was submitted by the hospital in accordance with this statute and its implementing regulations; the same was denied in a final decision entered by the Director of the Office of the Civilian Health and Medical Program of the Uniformed Services (“OCHAMPUS”) on October 16, 1987. This suit followed. On December 6, 1988, upon the parties’ joint motion, the case was remanded to OCHAM-PUS for reconsideration in light of new evidence and the agency’s expressed interest in reexamining the issues in the case. It is this revised final decision, dated April 19, 1989, (which reaffirms and incorporates the original decision by reference), that we now consider on appeal.
Before us are cross-motions for summary judgment raising two basic questions: first, may the regulations implementing the waiver standard be applied retroactively (as, in fact, they were in this case)? Second, was the recommended decision of the CHAMPUS hearing officer (adopted by and subsequently issued as the final decision of OCHAMPUS) erroneous as a matter of law and/or lacking in substantial evidence? Upon consideration of the parties’ briefs and the oral arguments offered in connection therewith, we decide the first issue in defendant’s favor, the second in plaintiff’s favor.
FACTS
The CHAMPUS beneficiary whose hospitalization expenses give rise to this case received inpatient psychiatric care at the McLean Hospital from October 24, 1984 until August 29, 1985 at which time he was discharged into a residential care center. The hospital received cost reimbursements from CHAMPUS (some pursuant to waivers) for the care provided through April 30, 1985; reimbursement for the care extended after that date was denied. Pursuant to the administrative procedures established for such cases, the question of the hospital’s entitlement to reimbursement based on waiver went before a CHAMPUS hearing officer who issued a decision recommending against allowance of the claim. The decision rested on the hearing officer’s conclusion that (i) the waiver criteria set forth at 32 CFR § 199.4(b)(5)(ix)(B) were valid and retroactively enforceable and (ii) the evidence offered by plaintiff had failed to meet all the criteria required for waiver as specified by the regulation. Specifically, after weighing the competing medical testimony presented at the hearing in light of the beneficiary’s clinical record, the hearing officer concluded that after May 1, 1985, the beneficiary, though still afflicted by an acute mental disorder, no longer presented a significant danger or risk either to himself or to others. It was therefore concluded that plaintiff had not met its burden of proof. In subsequent proceedings, this recommended decision was adopted as the agency’s final decision.
DISCUSSION
There are, as we have noted, two basic questions to be decided. The first, which we take up now, concerns plaintiff’s *154contention that its right to a waiver should have been determined under the general standard set forth in the waiver-authorizing statute—“extraordinary medical or psychological circumstances,” 10 U.S.C. § 1079(i)(4)—rather than under the regulations implementing that standard. The analysis offered in support of this result goes like this:
The regulation in question, 32 CFR § 199.4(b)(5)(ix)(B) (1987), although not promulgated until January 1986, some seven months after the time period involved in plaintiffs request, was nevertheless used as the benchmark to measure and ultimately to reject plaintiffs application for a waiver. It is this aspect of the regulation, namely, its retroactivity, that gives rise to the argument. Plaintiff maintains that, since the regulation was written pursuant to a specific delegation of authority to implement the statutory standard for the issuance of waivers, it is, in effect, a legislative or substantive regulation and, as such, may not be applied retroactively. Therefore, the argument continues, the only enforceable standard applicable to plaintiffs situation was that set out in the statute itself, namely, “extraordinary medical or psychological circumstances.” It is further argued that this general standard was satisfied in the case of the beneficiary in question, and, therefore, the denial of reimbursement for the hospital’s health care costs was contrary to law.
This argument cannot be accepted. It is not correct to say that rules promulgated by an agency pursuant to a specific delegation of authority, so-called “legislative” rules, may not be given retroactive effect. Granted, the Administrative Procedures Act, 5 U.S.C. §§ 551-554 (1988) defines a rule (i.e., a regulation) as “the whole or a part of an agency statement of general or particular applicability and future effect; ” id. § 551(4) (emphasis added). It is clear, however, that the courts have not demanded literal compliance with this definition. Rather, while rules of prospective application are the norm, those having a retroactive application are permitted in instances where the concerns for a coherent administrative policy are judged to outweigh the adverse effects of enforcing an after-the-fact standard.
The considerations that enter into this balancing approach are recited in Retail, Wholesale and Department Store Union v. N.L.R.B., 466 F.2d 380 (D.C.Cir.1972)—a case involving an agency’s attempt to apply retroactively a rule imposing back pay liability to redress a company hiring practice that was not unlawful (i.e., not an unfair labor practice) at the time the challenged practice took place. In declining to permit full retroactive application to the back pay sanction, the court identified the following factors as bearing on the issue in general:
(1) whether the particular case is one of first impression,
(2) whether the new rule represents an abrupt departure from well established practice or merely attempts to fill a void in an unsettled area of law,
(3) the extent to which the party against whom the new rule is applied relied on the former rule,
(4) the degree of the burden that a retroactive order would impose on a party, and
(5) the statutory interest in applying a new rule despite the reliance of a party on the old standard. 466 F.2d at 390.
Chiefly, the court was reluctant to burden the company by retroactively applying the rule because retroactive application was essentially unnecessary to further the statutory objective. To put it simply, enforcement would have been inequitable and without offsetting benefits to the administrative scheme.
There is nothing of that sort involved in this case. The regulation about which plaintiff complains announces no new standard for the granting of waivers. Rather, that regulation continues in place precisely the same criteria that existed as published CHAMPUS guidelines prior to the regulation’s promulgation.1 Consequently, plain*155tiff can make no claim that it relied to its detriment on a contrary rule. Therefore, the entirety of plaintiff’s argument comes down to the proposition that legislative rules may not be given retroactive application. And that argument, as the case law makes plain, is wrong. See e.g., Mason Gen. Hosp. v. Secretary of Dep’t of H.H.S., 809 F.2d 1220, 1224-28 (6th Cir.1987); Anderson, Clayton & Co. v. United States, 562 F.2d 972, 981 (5th Cir.1977); 2 Davis, Administrative Law Treatise § 7-23 (2d ed. 1979).
We come now to plaintiff’s second argument—that the case on the merits was wrongly decided. In order to qualify for a waiver of the 60-day limitation on the reimbursement of health care costs, the provider (plaintiff here) must satisfy two criteria. First, the provider must show that the patient is suffering “from an acute mental disorder or an acute exacerbation of a chronic mental disorder that results in the patient being put at significant risk to self or of becoming a danger to others.” 32 CFR § 199.4(b)(5)(ix)(B)(l) (1987). Second, it must be established that the patient requires “a type, level, and intensity of otherwise authorized service that can only be provided in an acute care inpatient setting.” Id.2
The hospital succeeded in establishing the second of these requirements but not the first. Stated simply, the hearing officer decided that while the patient’s behavior after April 30th continued to reflect instances of emotional disturbance (“significantly psychotic” are the words used), it was also true that he then exhibited “much improvement in cognitive functioning” and “increased ego development.” These two factors, according to the evidence relied on, were markers of increased stability and, correspondingly, of an individual who no longer posed a significant danger to himself or others. From this the hearing officer concluded that the hospital had failed to meet its burden of proof. Plaintiff challenges this conclusion saying that it is erroneous in law and without justification in fact. We take up these contentions now.
In deciding the case as he did, the hearing officer had to choose between two competing lines of testimony. On the one hand were the views of the treating psychiatrists—the doctors who had dealt with the patient on a daily basis and who urged his continued hospitalization on the ground that he remained a danger to himself and to others. On the other hand were reports of so-called “peer” reviewers—psychiatrists whose knowledge of the patient was drawn exclusively from the medical records. The hearing officer accepted the .opinion of the reviewing psychiatrists because, in his evaluation of the testimony, “the reviewing physicians’ findings were more adequately demonstrated in the medical records.”3 Plaintiff assails this conclu*156sion arguing that it embodies an incorrect understanding of the standard by which the evidence should have been judged.
Specifically, plaintiff contends that it is an established rule that, in evaluating opposing lines of medical testimony, the views of the treating physician must always be given greater weight than the views of other physicians who have had no personal involvement with the patient. Plaintiff further asserts that the opinion of a nonexamining doctor, standing alone, cannot constitute substantial evidence sufficient to overcome the contrary diagnosis of a treating physician. Plaintiff argues that the hearing officer ignored this rule, and that his decision is therefore wrong as a matter of law. Although we ultimately decide in plaintiff’s favor on the facts of the case, we do not agree with plaintiff’s view of the applicable law.
The rule to which plaintiff refers, the so-called “treating physician” rule, is a rule of common application in cases involving judicial review of administrative decisions denying claims for disability insurance benefits under the Social Security Act, 42 U.S.C. § 416(i) and § 423. See e.g., Piercy v. Bowen, 835 F.2d 190, 191 (8th Cir.1987); Hidalgo v. Bowen, 822 F.2d 294, 296 (2nd Cir.1987); Havas v. Bowen, 804 F.2d 783, 785 (2nd Cir.1986); Martin v. Secretary of Dep’t of Health, Educ. and Welfare, 492 F.2d 905 (4th Cir.1974). As typically expressed, the rule provides that:
a treating physician’s opinion on the subject of medical disability, i.e., diagnosis and nature and degree of impairment, is: (i) binding on the fact-finder unless contradicted by substantial evidence; and (ii) entitled to some extra weight because the treating physician is usually more familiar with a claimant’s medical condition than are other physicians, although resolution of genuine conflicts between the opinion of the treating physician, with its extra weight, and any substantial evidence to the contrary remains the responsibility of the fact-finder. Schis-ler v. Heckler, 787 F.2d 76, 81 (2nd Cir. 1986).
Thus, as the quoted text indicates, the treating physician rule does not give rise to an irrebuttable presumption in favor of the treating physician as plaintiff’s argument suggests. Rather, the rule holds only that the expert opinion of a claimant’s treating physician is “binding on the fact-finder unless controverted by substantial evidence.” Id. This rule, therefore, does no more than create a presumption of correctness in favor of the treating physician.
Similarly, we disagree with plaintiff’s assertion that the opinion of a nonexamining physician can never be regarded as substantial evidence. While it may be true that courts generally deem the opinion of a non-examining doctor insufficient to override the treating physician’s diagnosis, see, e.g. Havas v. Bowen, 804 F.2d 783, 786 (2nd Cir.1986); Martin v. Secretary of Dep’t of Health, Educ. & Welfare, 492 F.2d 905, 908 (4th Cir.1974), it is certainly not correct to say that such opinions are per se without probative worth. See Laws v. Celebrezze, 368 F.2d 640, 644 (4th Cir.1966). Rather, the weight to be given the opinion of a non-examining physician is to be judged in terms of the physician’s professional qualifications, the comprehensiveness of his knowledge of the patient’s medical history and the soundness of the reasons given in support of the opinion. In other words, to merit consideration the opinion of the non-examining physician must be an informed one.
That this requirement was met here cannot be doubted. The peer reviewer’s qualifications (Dr. Leslie Moldauer) were of a high professional order; her report demonstrates a thorough knowledge of the facts in the patient’s medical record; and her opinion as to the medical import of those facts, though differing from those of the *157treating physicians, was impliedly recognized to be disciplinarily sound. Clearly then, there was no case here for dismissing the peer reviewer’s opinion out-of-hand.
But to conclude that the peer reviewer’s opinion satisfies the criteria demanded of an informed judgment does not bring the analysis of this problem to an end. We must still determine whether the hearing officer’s decision is supported by substantial evidence. This requires an examination of the record as a whole, taking into account all the evidence supporting the decision as well as “whatever in the record fairly detracts from its weight.” Universal Camera Corp. v. NLRB, 340 U.S. 474, 487-88, 71 S.Ct. 456, 464, 95 L.Ed. 456 (1951). To withstand judicial review, the hearing officer’s decision must be reasonable in light of all the evidence in the record. United States v. Carlo Bianchi & Co., 373 U.S. 709, 715, 83 S.Ct. 1409, 1413, 10 L.Ed.2d 652 (1963).
The question before the hearing officer was which of two opposing medical diagnoses—the treating physician’s or the peer reviewer’s—better explained its position, facts and reasons considered. In deciding this case in accordance with the opinion of the peer reviewer, the hearing officer explained that he “looked to the hospital records in conjunction with the opinions to find indications which would substantiate each doctor’s conclusions and corroborate each doctor’s findings.” And, based on this review, he came to the conclusion that “the reviewing physicians’ findings were more adequately demonstrated in the medical records” or, as he also put it, “[t]he hospital records, in the present case, do not adequately demonstrate [that, after May 1, 1985, the patient presented a significant risk to himself or a danger to others].”
This language causes us some difficulty. To say, as the hearing officer did, that the reviewing physician’s findings were more adequately supported in the record is to address but half the question. Facts do not speak for themselves. Where the controversy is one between experts, it is the reasons given in favor of their views that needs to be examined. Thus, while there can be no doubt that the patient’s medical records exhibit those facts which the peer reviewer deemed significant to her conclusion—the absence of delusional behavior and overt violent threats, the capacity to verbalize anger and express fears—the real question here is whether those facts should have been considered controlling. Do they indicate that degree of stability and self-control which the peer reviewer attributes to them?
The treating physicians thought not. Though they were well aware of these aspects of the patient’s behavior—they, after all, had recorded them—the treating physicians discounted them as reliable signs of behavioral stability. It was their view that despite the higher levels of functioning and greater stability that the patient exhibited during the June through August period, he remained extremely vulnerable to perceived threats or losses and therefore stood at risk of reverting to the psychotic thinking and sexual aggressiveness that characterized his underlying illness.
Chief among the events that prompted this conclusion was an episode that occurred in early May (during the same period in question now) when the patient, then on a reduced regimen of anti-psychotic medication in response to promising signs of stability, reacted to a supposed injury to his father by blaming, and then by threatening to rape and to kill, a seven year old girl. In addition to this experience (which occurred during a home visit), other less severe acts of a regressive character took place later in the same month. These included making obscene phone calls, threatening female patients verbally and with gestures of lunging and kicking, and threatening the treating physician by taking her keys.
In light of these occurrences, the treating physicians believed that the signs of restabilization that again took hold in June through August (after drug dosages had been restored to initial levels) were not reliable indicators of self-control. That is, it could not be concluded that the patient would not present a significant danger to himself or others outside of the hospital’s *158structured environment. Indeed, one of the treating physicians diagnosed the patient’s condition in these terms: “under the surface of apparently unremarkable behavior during those months [July and August], he was still seething with the same kinds of rage and unbridled kind of sexual fantasies that were just being kept in control by very thin veneer.” OCHAMPUS Hearing Transcript, June 10, 1986 at 102-03.
Considering the evidence we have sketched, the question the court must decide is whether the hearing officer reasonably could have accepted the peer reviewer’s judgment in preference to that of the treating physicians’, as to the ultimate issue: whether the patient’s state of mind made him a significant danger to himself or others. Put another way, were the reasons given by the peer reviewer of sufficient probative integrity to stand as a reasonable counterweight to the views of the treating physicians? We think not. Three reasons prompt our conclusion.
First, as to the critical episode that occurred in early May (when the patient on a home visit threatened to rape and to kill a young girl), the peer reviewer, while acknowledging the regression demonstrated in this behavior, nevertheless considered it significant that the patient was able to verbalize that he acted out of anger. However, both the peer reviewer and the hearing officer failed to consider that this verbalization of reasons did not occur until two days after the event and then only after the patient had talked with his doctors about it. On the other hand, the treating physician, in light of these facts, saw in this episode little assurance of the patient’s capacity to arrest threatening behavior before it occurred. That seems to us the only reasonable conclusion to draw.
Second, as to the other events that took place later in that same month, the peer reviewer saw these as “threats done at quite a distance and many of them had a somewhat manipulative, gamey quality to them.” The treating physician strongly disagreed with this assessment. She acknowledged that while the patient, at times, was manipulative, “there [also] were other times, the ones that were cited here [referring to the above-mentioned events], in which he could be very frightening.” Here is her testimony:
... we see a wide range of adolescents [in the] inpatient unit, and ... some adolescents’ behavior is more manipulative. My experience with [the patient] is that he could be very frightening at times, he could be very frightening to me. And I took that seriously. I ... think there were times that he might have been manipulative, but I think that there were other times, the ones that were cited here, in which he could be very frightening. He would get, in a session like that, with either myself and occasionally with his mother, he would get extremely agitated. He would start to ... shake his feet and his hands and he would just sort of move closer and closer towards you and just sort of repeat the same thing over and over again, like “Right, right, I did this, right, right”—whatever it was. And there was very little way to kind of intervene or intercede or—or calm him down. And at those times the only way to calm him was to stop the meeting and to escort him to a quieter place. So I would very strongly disagree [with the peer reviewer’s] comments. I think what I’ve previously described ... went on in the beginning of May, both his calling from home to these girls, his threatening both the twelve- and seven-year-old on the unit—he would tend to kick and lunge towards both of them—his taking my keys and becoming threatening with me—I think ... all clearly, ... demonstrate how vulnerable and fragile [he was], and how without [his] being constrained in the hospital, ... he might have acted on ... these (inaudible) [presumably, impulses]. OCHAMPUS Hearing Transcript, June 10, 1986 at 76-77.
The hearing officer acknowledged the sharp difference of opinion that existed between peer reviewer and treating physician on the point in issue. The decision purports to resolve the conflict by saying: “[t]he overt threats should have been taken seriously; however, there were very few of them subsequent to May 1, 1985, and ac*159cording to the hospital records, they were quickly and quietly suppressed without any significant danger to the beneficiary or others.”
The analysis says more than is necessary; indeed more than the boundaries of the experts’ testimony would permit. The ultimate question to be answered was whether the patient represented a significant danger to himself or to others. Therefore, the matter in debate between the experts was the meaning of the patient’s upheavals, not the frequency of their occurrence or their amenability to control in a hospital environment. Simply put, the issue was whether the patient’s actions were ploys, masks to a nondangerous intent, or outbursts of rage demonstrating irrational thinking and a propensity to violence. The hearing officer’s conclusion that these actions were “serious” was as far as his analysis needed to go (or, logically, could have gone). In our view, no other conclusion could have been reached given that the treating physician had the opportunity to observe the patient and was, therefore, in a better position to form a judgment.
Third, and finally, we have this observation: given the acknowledged progress that the treating physicians had achieved in their care of the patient—a progress that the hearing officer described as “remarkable”—the matter in issue could only have been resolved in the hospital’s favor. Significantly, the administrative decision offers no explanation why the treating physicians’ diagnostic skills, so noteworthy in their accomplishments through April 30, 1985, should thereafter have gone off the mark. From a common-sense standpoint, the hospital’s record of achievement should stand as proof of the soundness of the judgment that, during the period from May through August 1985, the patient remained a significant danger to himself and to others. A decision that ignores a record of successful treatment in favor of a peer reviewer’s clinically untested diagnosis is not reasonable. This conclusion is especially pertinent here given the patient’s ea-pacity—as the peer reviewer put it—“to hide the severity of his thinking disorder,” and the assessment by one of the hospital’s doctors that he (the patient) “represented ... one of the five most difficult patients” encountered in thirteen years of professional experience.
In reaching our conclusion, we remain mindful of the concern that a court’s authority to review administrative decisions not be transformed into an authority to rewrite such decisions. However, we do not properly fulfill our role when the reviewing function becomes no more than a rubber stamp. We reverse here because, on the evidence presented, a reasonable decision would require a finding in favor of the hospital.
CONCLUSION
Based on the reasons stated, plaintiff’s motion for summary judgment is granted and defendant’s motion for summary judgment is denied. Plaintiff is entitled to judgment in the amount of $53,436 for inpatient medical care provided during the period from May 1, 1985 to August 29, 1985.4 The Clerk is directed to enter judgment accordingly.
. Prior to 1983, CHAMPUS paid for inpatient mental health care on an unlimited basis, as long as the care was medically necessary and at the appropriate level. This was changed by *155§ 785 of the Department of Defense Appropriation Act for Fiscal Year 1983, P.L. No. 97-377, 96 Stat. 1830, effective January 1, 1983, which limited funds for CHAMPUS cost-sharing of inpatient mental health care to 60 days per patient per year. However, § 785 also provided that the 60-day limit did not apply to services
provided pursuant to a waiver for medical or psychological necessities, granted in accordance with the findings of current peer review, as prescribed in guidelines established and promulgated by the Director, Office of the Civilian Health and Medical Program of the United States.
The Director, OCHAMPUS, issued the guidelines on December 29, 1982. Those guidelines contained the same criteria as were later incorporated into the regulation. CHAMPUS Policy Manual, Vol. I, Chap. I, § 12.
. Under subparagraph (B)(2) of the regulation, a waiver is also authorized where the providing of necessary mental health care on an outpatient basis is precluded because the individual is confined to a hospital due to a serious medical condition apart from his or her psychiatric condition. These conditions are of no concern in this case.
. Although there were two reviews conducted, each dealing with the period from May 1, 1985 to August 29, 1985, the hearing officer relied exclusively on the second of these reviews, i.e., a report prepared by Dr. Leslie Moldauer. As to the first peer review, conducted by Dr. William Barnum on May 17, 1985, we would agree with plaintiff that this report cannot constitute substantial evidence in this case because it was based on an incomplete medical record. In his review notes, Dr. Barnum frequently lamented the lack of information from which to form an opinion. The only information available to him were summary notes from March and progress notes from April; there was “[n]o documenta*156tion of any sort for the preceding 4 months." Such limited information would obviously impair the credibility of the report. Although the hearing officer’s decision describes the substance of Dr. Barnum’s views, and therefore speaks in terms of the opinions of the peer reviewers, it is clear that the hearing officer did not rely on the Barnum report. The analysis that the hearing officer accepted was the analysis provided by Dr. Moldauer. Hence, the inadequacies of Dr. Barnum’s report do not carry over to the hearing officer’s decision.
. Plaintiff also asks for attorney fees. However, we can see no basis for their allowance since the Government's litigating position was clearly a justifiable one.