OPINION OF THE COURT
BECKER, Circuit Judge.
I. Preliminary Statement
This case involves a claim by John Wier, a seventeen-year old mentally impaired boy from Monroeville, Pennsylvania, for Supplemental Security Income benefits based on disability. John appeals from a summary judgment in favor of the defendant Secretary of Health and Human Services upholding the Secretary’s denial of benefits. Although John was only eleven years old when his application for benefits was made by his mother, he will be well past eighteen, at which time he becomes subject to a new set of regulations,1 when his case is finally adjudicated properly. Once again we find that the failure of an administrative law judge to mention and explain medical evidence adverse to his position has deprived the Secretary of the substantial evidence necessary to sustain his determination. See Cotter v. Harris, 642 F.2d 700 (3d Cir.1981). We say “once again” because this is the fifth time since the beginning of the year that we have had to remand because of violation of Cotter, See Zelich v. Schweiker, 734 F.2d 9 (3d Cir.1984) (remanding for failure to consider medical evidence of treating physician); Muro v. Heckler, 729 F.2d 1448 (3d Cir. 1984); McIntosh v. Schweiker, No. 83-5463 (3d Cir. Feb. 28, 1984); Sanders v. Schweiker, No. 83-5360 (3d Cir. Feb. 29, 1984); see also Stokes v. Schweiker, 729 F.2d 932 (3d Cir.1984) (remanding case because of inadequate record); Sims v. Schweiker, 732 F.2d 147 (3d Cir.1984) (remanding because of reliance on stale evidence); Wallace v. Schweiker, 729 F.2d 1450 (3d Cir.1984) (remanding because of failure to consider impairments in combination); Speelman v. Schweiker, 727 F.2d 1101 (3d Cir.1984) (remanding because of general failure to present substantial evidence).
Ordinarily we would proceed at once to recount the factual and procedural history of this case. However, the procedural problems described in detail below impel us to highlight at the outset a broader problem, one that we believe to be an essential part of the universe in which this all-too-typical case should be seen. The problem may be stated simply: the system of adjudication created by Congress for resolving eligibility for disability benefits is not working as it should. Depending on one’s perspective, either the judicial system is placing impossible burdens on the administrative apparatus set in place by Congress to initially evaluate benefit claims or the officials in charge of the disability programs are, at least in many instances, ignoring the law. At all events, the resulting frictions and inefficiencies are burdening all concerned.
The following figures, compiled by the Administrative Office of the United States Courts and by the Clerk of the largest district court within this circuit, the District Court for the Eastern District of Pennsylvania, give some sense of the problem. To begin with, the number of suits filed by individuals denied disability benefits has exploded over the past several years. As [957]*957recently as 1979, only 866 suits were filed requesting the federal district courts to review disability adjudications. See 1983 Annual Report of the Director of the Administrative Office of the United States Courts 140 (Table 28). This number edged upwards to 976 in 1980 and then jumped to 1,628 in 1981, to 2,378 in 1982, and to 3,595 in 1983. Thus, the past three years have seen over a 400 percent increase in the number of filings. Figures from the Eastern District of Pennsylvania reflect the national experience. In 1982, 165 disability suits were filed. In 1983, this number doubled to 329. Preliminary figures for 1984 show the high level of filings to be continuing.
The problem is not so much the increase in filings and the consequent burden imposed on already-taxed federal judicial resources. Rather, what troubles us more is the fact that the Secretary so often appears to have wrongfully withheld benefits or to have processed a claim so that it is impossible to tell whether he correctly denied benefits. In the Eastern District of Pennsylvania, for example, over the past two and one quarter years, the Secretary has prevailed at the district court level only 31 percent of the time. This figure is extraordinarily low, not only because there is no other genre of cases with which we are familiar that is reversed or remanded by the federal courts at such a rate,2 but also because the Secretary does not even have to show that she is “right” but only that there is “substantial evidence” to support her position. Approximately. 45 percent of the cases are remanded to the Secretary for further proceedings. The claimant wins outright or subject to further proceedings in 22 percent of the cases initially filed.3
In terms of appellate cases, although we do not have exact figures either nationally or on a local level, it appears as though, even after the apparently rigorous district court screening, the courts of appeals reverse district court judgments in favor of the Secretary approximately one quarter of the time. See 1983 Annual Report of the Director of the Administrative Office of the United States Courts 224 (Table B-1A). This figure is apparently on the rise, for since the beginning of the year, the reversal rate by the Third Circuit has been about fifty percent. These figures showing the resolution of cases presented to the district court probably overestimate the number of times that the Secretary has substantial evidence to support her position, for our figures of course cannot take into account the number of claimants who cannot ask for judicial review of the Secretary’s determination or take an appeal from an adverse district court judgment because their financial resources have been exhausted by the lengthy and expensive process of seeking benefits. Neither are statistics available to measure the delays between initial application and final disposition, but our judicial perspective tells us that they are often considerable.
It is not our province to say whether the delays and injustices reflected in these statistics are the consequence of judicial persnicketiness, of an executive branch attempting to reduce governmental expenditures, or of other factors. It does seem clear, however, that the system is not working to anyone’s advantage and cries out for Congressional attention.4 We now [958]*958turn to the case before us, which, in our opinion, provides a more human illustration of the problems reflected in the statistics and which also presents important questions concerning analysis of children’s impairments in combination and the power of the administrative law judge to make unexplained credibility determinations.
II. PROCEDURAL HISTORY
John Wier applied through his mother Janet for SSI disability benefits on August 10, 1978. Administrative levels of the Social Security Administration denied his claim. John requested a hearing, which was held nine months after benefits had been requested. The Administrative Law Judge (AU) decided two months later that John was not disabled under the applicable statutes and regulations. On August 31, 1979, a year after John first requested benefits, the Appeals Council of the Social Security Administration denied review.
John then petitioned the United States District Court for the Western District of Pennsylvania to review the determination of the Secretary that he was not entitled to SSI benefits. The court referred cross motions for summary judgment to a magistrate for a recommendation. After an initial examination of the evidence before the Secretary, the magistrate recommended that the Secretary’s decision be affirmed. The magistrate issued this recommendation, however, without benefit of briefing from the appellant, whose counsel at the time had been delinquent in his submissions. On April 24, 1980, upon receipt of briefs from appellant’s attorney, and reports detailing the results of additional psychological evaluations, the magistrate revised his opinion. In his Supplemental Report and Recommendation he wrote:
Both additional reports contain evaluations of the plaintiff which indicate that although his I.Q. exceeds the maximum requirement [sic] for a finding of disability.... [T]here appears to be overriding considerations which indicate that the plaintiff is functioning considerably below the low level upon which one would expect him to function as a result of his test scores. Thus there may be a severe overlay or interacting of factors which when combined with his limited intelligence would warrant a finding of disability.
The magistrate thus recommended a remand for consideration of this new evidence submitted by appellant. The district court followed the magistrate’s recommendation and on May 15, 1980, remanded the case to the Social Security Administration.
Given the rather simple directive óf the magistrate, and given the obvious need for haste if John was to be benefitted by disability payments while still a child, one might think the Social Security Administration would have resolved the remanded matter.in several weeks or months. Alas. It took the Administration’s Appeals Council two months to determine that, in light of the district court’s disposition, further fact finding was needed. After another four months, in December 1980, a hearing was held before an administrative law judge to examine the new evidence, which consisted of testimony from several witnesses and three medical reports. In January 1981, the administrative law judge issued a recommended decision. Five months later, in May 1981, the Appeals Council discovered that the testimony offered at the hearing held in December 1980, had not been properly recorded. The Appeals Council thus remanded the case back to the administrative law judge for proceedings “consistent with the Order of the District Court and with the Appeals Council Order of Remand dated July 8, 1980.” The Appeals Council also directed the administrative law judge to “obtain a consultative psychiatric examination with a functional capacities evaluation and with psychological testing.”
One might also have thought, given the Social Security Administration’s own culpability in the delay thus far, and given the [959]*959rapidly waning prospects for John’s receiving aid while still a child and capable of rehabilitation, that the case would have been expedited. Instead, no hearing was held for almost eight months, at which point three medical reports and one school report were submitted. After another two month delay, in March 1982, the ALJ issued another recommendation that no benefits be awarded. The Appeals Council denied review in June 1982; this constituted the final action of the Secretary.
With the Social Security Administration at last having processed the directive of the district court, the case returned to the judicial system. The district court again referred cross motions for summary judgment to a magistrate. In March, 1983, the magistrate recommended that defendant’s motion for summary judgment be granted. The district court approved this recommendation.
John Wier, now 17½ years old, thus stands before us on appeal and asks us to reverse the judgment of the district court.
III. REGULATORY FRAMEWORK
Under 42 U.S.C. § 1382c(a)(3)(A) (1976), an individual is considered disabled (and thus eligible under certain other conditions for benefits) “if he is unable to engage in any substantial gainful activity by reason of any medically determinable physical or mental impairment which can be expected to result in death or which has lasted or can be expected to last for a continuous period of not less than twelve months (or in the case of a child under the age of 18, if he suffers from any medical determinable physical or mental impairment of comparable severity).” (Emphasis added). The Secretary has implemented this provision through 20 C.F.R. § 416.923 and 20 C.F.R. § 416.926. Section 416.923 states:
We will find that a child under age 18 is disabled if he or she—
(a) Is not doing any substantial gainful activity; and
(b) Has a medically determinable physical or mental impairment(s) which compares in severity to any impairments) which would make an adult (a person age 18 or over) disabled. This requirement will be met when the impairment(s)—
(1) Meets the duration requirement; and
(2) Is listed in Appendix 1 of Subchapter P of Part 404 of this chapter; or
(3) Is determined by us to be medically equal to an impairment listed in Appendix 1 of Subpart P of Part 404 of this chapter.
The meaning of medical equivalence as described by 20 C.F.R. § 416.923(b)(3) is further elaborated in 20 C.F.R. § 416.926, which states in part:
(a) How medical equivalence is determined. We will decide that your impairments) is medically equivalent to a listed impairment in Appendix 1 of Subpart P of Part 404 of this chapter if the medical findings are at least equal in severity and duration to the listed findings. We will compare the symptoms, signs, and laboratory findings about your impairments), as shown in the medical evidence we have about your claim, with the medical criteria shown with the listed impairment. If your impairment is not listed, we will consider the listed impairment most like your impairment to decide whether your impairment is medically equal. If you have more than one impairment, and none of them meets or equals a listed impairment, we will review the symptoms, signs, and laboratory findings about your impairments to determine whether the combination of your impairments is medically equal to any listed impairment.
(b) Medical equivalence must be based on medical findings. We will always base our decision about whether your impairment(s) is medically equal to a listed impairment on medical evidence only. Any medical findings in the evidence must be supported by medically acceptable clinical and laboratory diagnostic techniques. We will also consider the medical opinion given by one or more physicians designated by the Secretary in deciding medical equivalence.
[960]*960Appellant has four theories under which he claims to be entitled to benefits under the statute and regulations. First, he claims that he is mentally retarded within the meaning stated in Appendix 1. Second, he claims that he is suffering a functional non-psychotic disorder within the meaning stated in Appendix 1. Third, he claims that he has a “communication impairment, associated with a documented neurological disorder” within the meaning stated in Appendix 1. Fourth, appellant claims that he is disabled under 20 C.F.R. § 416.923(b)(3) and 20 C.F.R. § 416.926 because his combination of impairments are “medically equal” to one of the listed impairments. We consider each of these theories in turn.
IV. THE IMPAIRMENTS
A. Mental Retardation
Under section 112.05 to Appendix 1 of Subchapter P, a person under age 18 suffers from mental retardation if there is: “A. Achievement of only those developmental milestones generally acquired by children no more than one-half the child’s chronological age; or B. IQ of 59 or less; or C. IQ of 60-69 inclusive, and a physical or other mental impairment imposing additional and significant restriction of function or developmental progression.”
Appellant admits that, with IQ scores ranging from the seventies to the eighties, he cannot literally meet any of the requirements for mental retardation. Based upon the “medically equal” language of 20 C.F.R. § 416.923(b)(3), he argues, however, that his functional ability is that of a person with an IQ in the range specified in the regulations. The Secretary’s short answer to this “functional ability” contention is that the appellant is essentially making a “developmental milestones” argument, which is explicitly prohibited by the SSI regulations where, as here, the claimant has taken intelligence tests. The Secretary is correct on this point. The general language of 20 C.F.R. § 416.923(b)(3) permitting a finding of disability based on medical equivalence is overridden by the specific language of section 112.00B, which states that “Developmental milestone criteria may be the sole basis for adjudication only in cases where the child’s young age/or condition preclude formal standardized testing by a psychologist or psychiatrist experienced in testing children.”
Appellant makes a second argument that, in fact, he has scored in the 60s on the Wide Range Achievement Test and that this low score, coupled with other physical or mental impairments, entitles him to benefits on grounds that his condition is medically equal to that stated in section 112.05C. The Secretary responds, again correctly, that, under the regulations, achievement tests results are not to be used where normal intelligence tests are available. The second paragraph of section 112.00B of the listing of impairments, states that “Standardized intelligence tests, such as the Wechsler Preschool and Primary Scale of Intelligence (WPPSI), the Wechsler Intelligence Scale for Children (WISC), the Revised Stanford-Binet Scale, and the McCarthey Scales of Children’s Abilities, should be used whenever possible.” John has scored in the 70s on the WISC. Accordingly, he cannot receive benefits based upon his low intelligence alone.
B. Functional Non-Psychotic Disorder
Section 112.04 of Appendix 1 of Subchapter P lists as an impairment a “Functional nonpsychotie disorder”. This disorder is “Documented by psychiatric evaluation and supported, if necessary, by the results of appropriate standardized psychological tests and manifested by marked restriction in the performance of daily age-appropriate activities; constrictions of age-appropriate interests; deficiency of age-appropriate self-care skills; and impaired ability to relate to others; together with persistence of one (or more) of the following: ... (G) Asocial or antisocial behavior.”
Appellant, relying primarily on the testimony of his mother, argues that he is entitled to benefits on grounds that his asocial and antisocial behavior satisfies section 112.04(G). Janet Wier testified that John has no friends and that he spends his time watching television (with little comprehen[961]*961sion). She further testified to numerous instances of antisocial behavior by John such as urinating out a window of her house, cutting down a neighbor’s flowers while mowing her lawn, and disregarding rules set forth by his parents and teachers. The Secretary responds to this argument by stating that recovery on this ground is foreclosed because the administrative law judge expressly discredited the testimony of appellant’s mother and that, “inasmuch as the Administrative Law Judge has the opportunity to observe demeanor and determine credibility, his observations on these matters must be given great weight.” In support of this proposition the Secretary cites two district court cases, Good v. Weinberger, 389 F.Supp. 350 (W.D.Pa.1975) and Gardner v. Richardson, 383 F.Supp. 1 (E.D.Pa.1974). The appellant rejoins by citing (accurately) Davidson v. Harris, 502 F.Supp. 1208 (E.D.Pa.1980), for the proposition that “an administrative law judge may not reject a claimant’s testimony without some evidence in the record to support this rejection.”
The question as to how much weight to give to the credibility findings of administrative law judges strikes at some of the central issues in the burgeoning area of disability law. Unlike Article III judges, the administrative law judges implementing the various social security benefits programs do not have lifetime tenure and may be subject to bureaucratic and other pressures. The frequency with which the federal courts have reversed determinations of administrative law judges in this area, see supra part I, particularly in recent years, may to some degree reflect their beliefs about the magnitude of these pressures. For whatever reason, there is much in our circuit’s jurisprudence (and that of other circuits) that requires administrative law judges in social security cases to explicate the analysis upon which they base their ultimate decisions, see generally Cotter v. Harris, 642 F.2d 700 (3d Cir.1981), especially where they are rejecting the testimony of the claimant’s treating physician or allegations (particularly those supported by the treating physician) of intractable pain. We cannot gainsay that the requirements of Cotter and the concerns underlying these cases might be evaded by lawless or pressured administrative law judges if they were able simply to state that they disbelieved witnesses who gave testimony contrary to their preconceived positions. See Baerga v. Richardson, 500 F.2d 309 (3d Cir.1974), cert. denied, 420 U.S. 931, 95 S.Ct. 1133, 43 L.Ed.2d 403 (1975).
Against these arguments favoring detailed findings, we must recognize that most ALJs are able and conscientious judicial officers, and that there may be systemic problems in administrative programs that cannot be eliminated by judicial requirements that the ALJ expand and elaborate upon the record presented to reviewing courts. Specifically, we doubt that any problems with administration of the social security laws would be significantly improved by requiring ALJs to articulate the factual findings undergirding all credibility determinations. To begin with, an ALJ faithful to his statutory mandate might well be unable to verbalize the basis on which he rejected testimony as lacking credibility. Belief or disbelief of a witness is often based on a trier of fact’s subliminal aggregation of miniature instances of witness behavior that cannot be individually recalled or described. Second, even as to those instances of behavior that are susceptible to verbal description, we doubt that their enumeration by the ALJ would enhance the ability of a tribunal (such as this) reviewing a cold record to assure fidelity to law. Realistically, findings such as “the witness shifted in his chair” or “the witness failed to look his questioner in the eye” simply cannot be challenged on review. In short, a requirement that an administrative law judge set forth the factual findings undergirding his credibility judgments would have value only in those instances where those findings are, by reason of other facets of the record, see, e.g., Zelich v. Schweiker, 734 F.2d 9 (3d Cir.1984); Daring v. Heckler, 727 F.2d 64 (3d Cir.1984); Wallace v. Secretary of HHS, 722 F.2d 1150, 1155 (3d Cir.1983); Dobro[962]*962wolsky v. Califano, 606 F.2d 403 (3d Cir.1979), susceptible to judicial review, or where the process of making findings focuses the attention of the administrative law judge more sharply on the evidence before him. The resulting increase in faithful execution of the social security laws should not, of course, be ignored; but neither should courts overstate the limited utility of such findings or underestimate the burden or indignity their recitation would impose on already-beleaguered administrative law judges.5
Fortunately, we do not have to decide in this case whether an administrative law judge’s failure to give any explanation of his credibility judgments constitutes grounds for remand, for in this case the administrative law judge gave a reason. Hence we need only decide whether it is enough in itself that he gave a reason or that the reason is sufficient. The administrative law judge noted that on at least one occasion, the mother’s testimony was seriously at odds with objective evidence before him. Mrs. Wier testified that “the claimant’s legs had been run over by a fire truck.” “If the claimant’s legs had been run over by a fire truck,” the administrative law judge noted, “there would be no reason for pursuing this claim based on a mental disability.” Moreover, the appellant produced virtually no other evidence to demonstrate a functional non-psychotic disorder, and the administrative law judge had other evidence before him, including reports from doctors, saying that the appellant could be “pleasant and cooperative,” (at least with them). Under these circumstances, we believe the administrative law judge satisfied whatever requirement there may be to state the reasons underlying credibility judgments.6 Accordingly, we do not believe the appellant has sustained his burden of showing disability based on a functional non-psychotic disorder. See Hess v. Secretary of Health, Education & Welfare, 497 F.2d 837 (3d Cir.1974).
C. Communications Impairment
The third impairment claimed by appellant as the basis for a disability finding is listed in section 111.09 of Appendix 1: a “communication impairment, associated with documented neurological disorder and ... [a] documented speech deficit which significantly affects the clarity and content of the speech ....” The Secretary concedes that appellant suffers a neurological disorder (minimal brain damage) and concedes that the appellant has difficulty in communicating. The real dispute between the parties is whether his “speech deficit” is sufficiently significant.
In support of his contention that the appellant’s speech disorder is not significant, the Secretary relies upon a report from Dr. Hiram Wachs, a neurologist, who examined appellant in 1980. In his brief report, Dr. Wachs states that appellant’s “speech is understandable.” The Secretary also relies on an unsigned (and almost unreadable) “individual educational plan” progress report from appellant’s school. This report states that appellant has “good speech intelligibility with adequate voice fluency for communication through simple and generally complete sentences.” The ALJ found this evidence dispositive.
What the ALJ did not mention in his report, however, was that two doctors came to conclusions apparently different [963]*963than those of Dr. Wachs. Dr. Lee Barolo, a psychologist who examined appellant on several occasions, reported in 1981 that appellant’s speech was “10 to 15% unintelligible” and that appellant “evidenced immature sound substitutions.” Dr. Lloyd Bell, a psychiatrist who examined appellant, reported that appellant’s speech was “significantly distorted.” The ALJ’s “failure to explain his implicit rejection of this evidence or even to acknowledge its presence” violated this court’s directive in Cotter v. Harris, 642 F.2d 700, 707 (3d Cir.1981). Accordingly, we must remand this ease for consideration in light of Cotter.
D. Combination of Impairments
Finally, the appellant argues that even if his impairments in intelligence, speech and behavior do not quite measure up to the standards required by the listings, taken in combination their severity is equivalent to that of impairments set forth in the listings. Thus, appellant argues, under 20 C.F.R. § 416.923(b)(3) and 20 C.F.R. § 416.-926, he is entitled to disability benefits. The Secretary, while conceding that a combination of impairments can entitle an otherwise ineligible minor to disability benefits, argues that the evidence before the ALJ sufficiently demonstrated that appellant’s impairments, even taken in combination, did not render him disabled. Appellant rejoins that the evidence cited by the Secretary was deficient.
The regulations promulgated by the Secretary oblige her to consider in combination the impairments of applicants for disability benefits. In the case of children, this obligation is a difficult one to fulfill.7 Implicit in the combination regulations is the notion that all of a child’s diverse medical problems have some quantifiable level of “severity” and that mathematical processes can combine the “severity level” of the listed impairments. The problem, of course, is that medical problems are largely incommensurables; the Secretary’s regulations require him to perform the arguably impossible (or meaningless) task of adding apples and oranges. See generally J. Mashaw, Bureaucratic Justice, 109-14 (1983). The statutes passed by Congress require the courts to see whether the Secretary has appropriately carried out this task.
In this case, the Secretary relies on brief reports of three doctors who never saw the appellant, but only reviewed reports. They declare that the appellant’s impairments do not meet or equal the listings. Dr. Enyeart states “His impairment does not meet or equal the listings. The claim is denied.” The report of Dr. Usyk checks a box stating “claimant not disabled through date of current determination.” Dr. Benton states, “In conclusion, although it is clear that this child has significant handicaps, in my opinion they are not sufficiently severe as to meet or be the equivalent of those cited in the applicable Social Security regulations.”
An initial problem with the Secretary’s reliance on the three reports is that the doctors who issued them never saw John Wier, but relied solely on the record placed before them. Their opinions consequently have less probative force, as a general matter, than they would have had if they had treated or examined him. See Cotter v. Harris, 642 F.2d 700, 704 (3d Cir.1981). Aside from the general problems inherent in reliance upon the opinions of non-examining physicians, the Secretary’s reliance on those opinions is particularly troubling in this case because the three reports are [964]*964extraordinarily stale. Two of the reports are dated from 1978, when appellant was 11. The third report, a brief letter, dates from 1980, when appellant was 13. In a case involving an adolescent, where medical and psychological problems often change rapidly, reliance on this sort of evidence is highly suspect. The staleness of the reports cannot be measured by their age alone, however. Since these reports were issued, Drs. Bell and Barolo have concluded — albeit without recognition by the administrative law judge, see supra p. 963 that appellant has a serious speech problem. There is also evidence in the record that would suggest that appellant’s mental development relative to his chronological age has slowed in adolescence. Without this late data, the reports of the three doctors are virtually worthless.8
Because of these serious problems with the Secretary’s finding on combination of impairments and because this case must be remanded anyway to consider our directive in Cotter v. Harris, we believe that the record should remain open for appellant to present evidence and argument that appellant’s impairments, taken in combination, rise to the level of severity required by the listings in Appendix 1.9 The Secretary may likewise present evidence on this point.
The judgment of the district court will be reversed and the case remanded for further proceedings consistent with this opinion.