Bernard E. Miller appeals the denial of his application for social security disability benefits. Finding no substantial evidence in the record to support the administrative law judge's conclusion that Miller had a residual functional capacity for the full range of light or sedentary work, we reverse and remand with directions to award benefits.

* This case has now been pending in the Department of Health and Human Services or in the federal courts for over six years. Miller filed his original application for disability benefits on December 14, 1982. He alleged a number of exertional and nonexertional impairments, including a back injury, severe headaches, leg pain, loss of hearing and eyesight, a hernia, ulcers and severe chest pain. At each of the three administrative hearings held in the case, Miller also presented evidence that he suffered from various nonexertional psychiatric impairments, including chronic anxiety and depression.

The first administrative law judge to consider these claims concluded that Miller was unable to perform his past relevant work as a mechanic and welder. The ALJ found, however, that the claimant nevertheless had a "residual functional capacity for the full range of light work." Joint Appendix at 11. The Appeals Council denied Miller's request for review. The district court remanded the case for further proceedings, however, largely on the basis of a magistrate's report finding that the ALJ had failed adequately to consider certain evidence "clearly establish[ing] the existence of nonexertional impairments, [including] pain ..., dizziness ..., and anxiety and depression." Id. at 19. In the face of such a showing, "the Secretary is required to produce evidence in the form of expert vocational testimony which would support a finding that, despite plaintiff's 'combination of nonexertional and exertional impairments, specific jobs exist in the national economy which he can perform.' " Id. at 19-20 (quoting Grant v. Schweiker, 699 F.2d 189, 192 (4th Cir.1983)). The Department had failed to present such evidence, and Miller was therefore entitled to a supplemental hearing.

On remand, a second ALJ heard testimony from a vocational expert and concluded that there were indeed "a significant number of jobs in the national economy which [the claimant] could perform." Id. at 35. He in turn recommended that the Secretary once again find Miller not disabled. On Miller's renewed request for review, however, the Appeals Council reversed and remanded for a second supplemental hearing, finding that the claimant had been prejudiced by an improper computation of the period during which he was potentially eligible for benefits. The ALJ had found preliminarily that Miller was last eligible for disability benefits on September 30, 1984, thereby precluding the introduction or consideration of any evidence suggesting that he became disabled after that date. On the basis of an updated earnings record, however, the Appeals Council concluded that Miller's disability coverage extended through September 30, 1986. The claimant was entitled, therefore, to a "redetermination as to [his alleged] disability" during this two-year, "unadjudicated" period. Id. at 49. The Council was apparently concerned, moreover, that the Department still had failed to give adequate weight to the evidence of Miller's nonexertional impairments, including his "chronic, prolonged headaches ..., withdrawal from situations with loud noises and crowds, generalized lack of stamina, and probable difficulty in meeting the production standards of competitors and sustained work." Id. at 50. It instructed the ALJ to reconsider explicitly on remand "the effects of nonexertional limitations and pain on the claimant's residual functional capacity, any side effects of medications, and the availability of jobs for the claimant." Id.

The Department's regional administrator assigned the case to yet another ALJ, who conducted a de novo evidentiary hearing in March of 1987. She ultimately recommended renewal of the Secretary's initial determination that Miller was not disabled, finding that the claimant had an "exertional capacity for both light and sedentary work," and that his "nonexertional limitations, [which are] insufficient to rise to the level of nonexertional impairments, in no way ... compromise his capacity for [such] work." Id. at 66. On Miller's third and final request for review, the Appeals Council again readjusted the expiration date of the claimant's benefits eligibility period, concluding on the basis of further updated earnings records that he last met the special earnings requirement on December 31, 1986. The Council nevertheless found, in agreement with the ALJ's ultimate conclusion, that Miller had a residual functional capacity for gainful employment. It therefore denied disability benefits. The district court affirmed, and this appeal followed.

We review the record, as did the district court, to determine whether there was substantial evidence to support the Secretary's conclusion that the claimant was not disabled at any time before his eligibility for benefits expired. 42 U.S.C. Sec. 405(g). See also Richardson v. Perales, 402 U.S. 389, 401 (1971); Lester v. Schweiker, 683 F.2d 838, 841 (4th Cir.1982). In so doing, we must remain mindful that the ALJ here found the claimant incapable of performing his past relevant work and therefore reached the fifth and final step of the sequential evaluation process. At this point, the Secretary assumed the burden of coming forward with competent evidence that, notwithstanding his severe impairments, Miller had a residual functional capacity for the tasks required by jobs then available in the national economy. Hall v. Harris, 658 F.2d 260, 264 (4th Cir.1981) (citing McLamore v. Weinberger, 538 F.2d 572, 574 (4th Cir.1976) and Taylor v. Weinberger, 512 F.2d 664, 666 (4th Cir.1975)). Here, we are persuaded both that the Secretary failed to carry this burden and that, as a general matter, there is no substantial evidence in the record to support the conclusion that the claimant had a residual functional capacity for gainful employment.

If the ALJ's consideration of this claim could properly have been confined solely to the evidence of exertional impairments as they affected Miller's residual functional capacity, the ultimate finding that he was not disabled would be substantially supported thereby. That evidence consists primarily of various medical reports suggesting that the claimant suffered from a number of orthopedic and gastrointestinal difficulties. The record indicates that, over a period of six years, Miller sought treatment from no less than ten physicians for a wide range of physical injuries and ailments. Several of these doctors concluded that the claimant was incapable of returning to his former position as a truck mechanic, but it was never suggested that Miller's physical condition had deteriorated such that other, less strenuous work would be impossible. After treating the claimant for a number of orthopedic injuries, for example, Dr. Elwood H. Heilman wrote to officials of the West Virginia Workmen's Compensation Fund and reported that "this patient has reached his maximum degree of improvement. I don't think he needs further treatment. I think he has a 10% partial permanent disability as a result of [his back] injury and any other injuries he has had." In January of 1983, Dr. H.M. Mills examined Miller and found that the claimant's generalized orthopedic distress was "not a serious situation." Significantly, nothing in the record suggests that the claimant's joint disease or back injury had worsened markedly by the time his eligibility for disability benefits expired in December of 1986.

The resulting inference that Miller was not disabled as a result of exertional impairments finds further support in the reports of the two physicians who conducted comprehensive physical examinations of the claimant. Dr. Bruce Guberman examined Miller in January of 1983 and found evidence of degenerative joint disease, traumatic spinal arthritis, peptic ulcers and a slight sensori-neural hearing loss. He concluded, however, that the claimant's combined physical impairments threatened "no restrictions of daily activities." Similarly, Dr. J. Loren Smith examined Miller in April of 1984 and concluded that the "prognosis is favorable for this patient to return at some time in the future to employment, possibly related to a skill not requiring heavy physical exertion."

On the basis of this and similar evidence, much of which suggests that Miller's physicians treated his various physical impairments conservatively and without prescription drugs, we would have to conclude that substantial evidence supports the Secretary's conclusion that the claimant's exertional impairments standing alone did not result in compensable disability.

But that of course does not end the matter. As both the Appeals Council and the district court had recognized on earlier administrative and judicial reviews of benefits denials, there was also in this record significant evidence of nonexertional impairments that should have been taken into account in assessing Miller's claim. The law clearly requires that the Secretary consider the combined effects of exertional and nonexertional impairments, see Grant, 699 F.2d at 192, and both reviewing bodies were sufficiently concerned that this had not properly been done that they remanded to insure that it would be. Though on the third round the Appeals Council again recognized a possible failure by the ALJ to have considered the matter properly, it decided that the final determination of nondisability was now supported by substantial evidence.

We disagree. We conclude to the contrary that once again and in a way that makes her nondisability determination untenable, the third ALJ failed properly to assess the evidence of nonexertional limitations on this claimant's ability to engage in gainful employment.

Specifically, we are satisfied that the ALJ failed properly to assess the evidence that Miller suffered from such severe anxiety and depression that normal functioning in the workplace had become an impossibility for him during the covered period of his employment. Indeed we conclude that the relevant medical evidence not only strongly corroborates this claim, but stands uncontradicted in the record.

That evidence discloses that Miller visited psychiatric clinics for disability evaluations on three separate occasions. On September 23, 1983, Dr. Michael McNeer conducted a comprehensive examination and noted that Miller was "markedly anxious" and "noticeably depressed." Dr. McNeer diagnosed moderately severe chronic generalized anxiety disorder, mild to moderate chronic dysthymic disorder, and "psychological factor[s] affecting physical condition." He noted that Miller engaged in a relatively high level of daily activity, including some fishing, hunting and church work, but concluded that the claimant's

chronic close to severe anxiety, depression, chronic back pain, peptic ulcer disease, and chest pain constitute a severe impairment for returning to substantial, gainful employment at this time when considered in light of his age, education, background, and work experience.

Record Transcript at 230.

Dr. McNeer's clinical psychologist, William Brezinski, conducted a separate examination and also found evidence of disabling nonexertional impairments. He concluded that Miller was "chronically anxious" and "presently functioning in the lower portions of the average range of intelligence." The claimant appeared to be "unassertive, somewhat inadequate, and at times unable to cope with everyday stress and responsibility." He exhibited "psychological symptomotology of a wide variety," and "his symptoms [would] likely ... intensify significantly during times of stress."

Miller sought an updated evaluation from the Charleston Psychiatric Group in February of 1986. Psychiatrist Dr. Ralph S. Smith noted that Miller was taking Flexeril, Librium, Felderic and Nadelate, but diagnosed no significant impairments apart from "psychological factors affecting physical disease." Psychologist Dr. Teresa D. Smith conducted a separate examination, however, and found poor basic adjustment, anxiety, depression, paranoia and schizoid tendencies. She concluded that Miller's "ability to go to work on a sustained basis is questionable." His "capacity to carry out instructions under ordinary supervision is marginal," and "it would be difficult for [him] to meet production norms occupationally."

Eleven months later, the claimant's psychiatric condition had apparently worsened. In January of 1987, he visited the Saint Albans Mental Health Services Center in Beckley, West Virginia. A board-certified psychiatrist and neurologist, Dr. Ramseh Shah, and two clinical psychologists, Crystal Whittington and Gregory Bowland, conducted separate examinations. They diagnosed moderately severe to severe chronic generalized anxiety disorder, severe chronic dysthymic disorder, psychological factors affecting physical condition, moderately severe "psycho-social stressors," and poor adaptive functioning. Their report concluded as follows:

We recommend a psychiatric permanent partial disability of 25%.... We do not believe that [Miller] will be able to return to a work environment which requires lifting, bending, and twisting and other strenuous work. He is likely to be considered a poor candidate for vocational rehabilitation due to limited education, age, and physical limitations. He would appear then, to be totally and permanently disabled for participation in gainful competitive employment.... He was advised to seek psychiatric treatment. The prognosis is poor.

Id. at 467 (emphasis supplied). The record does not reveal whether Miller indeed sought further treatment, or whether his condition changed after the date of the Saint Albans report. Of course, the Department conducted the final administrative hearing in this case only two months later. We find it significant, however, that Dr. Shah's findings are not contradicted by any evidence of record, and that they indeed find support in the earlier reports of other specialists.

Taken as a whole, therefore, the record in this case reveals a consensus among Miller's various treating physicians and the consultive psychologists that he suffered from nonexertional psychiatric impairments of sufficient severity to preclude normal functioning in the workplace. Conceding the presence only of "mild impairments in the claimant's mental functioning," however, the ALJ discredited most of the relevant medical evidence and rejected Miller's effort to establish a conjunction of exertional and nonexertional limitations causing disability. See J.A. at 62-63. Looking carefully to her stated reasons for doing so, we find them neither logically persuasive nor legally sufficient.

First, the ALJ placed repeated emphasis on those portions of the treating physicians' reports that included notations of Miller's "fairly full and independent lifestyle.... The claimant is active in church, hunts and fishes, drives, told psychiatrist Smith he went shopping, played cards with friends, watched television, and worked with video equipment." Id. at 61. The ALJ apparently considered the evidence of these activities inconsistent with the professionals' conclusions that Miller suffered from "significant anxiety or depression." Id. We are not persuaded, however, that proof of an ability to carry on "normal daily activities" is sufficient, standing alone, to overcome otherwise uncontradicted medical opinions that a claimant is occupationally disabled as the effective result of nonexertional impairments. The evidence here strongly suggests that Miller's limitations are stress-related, hence particularly acute in settings involving even minimal supervision or the assumption of limited responsibility. Psychologist Brezinski opined, for example, that Miller's "symptoms are likely to intensify significantly during times of stress." It would appear, moreover, that the treating physicians themselves found no inherent inconsistency between the evidence of somewhat normal daily functioning and anxiety-induced occupational disability. Dr. McNeer specifically concluded that Miller suffered from a "severe impairment for returning to substantial, gainful employment" notwithstanding a "high level of activity." The point is that psychiatric impairments might make it impossible, as the specialists here opined, for a patient to cope with the everyday stresses associated with a steady job, even though they do not prevent functioning in less stressful circumstances. Thus where, as here, the unvarying medical consensus among examining and treating professionals is that the claimant suffers from disabling, nonexertional impairments, evidence of normal activity in non-occupational settings simply is not sufficient to support a finding of residual functional capacity for substantial gainful activity.

Second, the ALJ relied heavily--and we think improperly--on the opinion of a consultive reviewing psychiatrist, Dr. Charles Weise, who did not examine the claimant and instead based his testimony on a review of Miller's psychiatric and psychological records. In response to direct questioning from the ALJ, Dr. Weise expressed considerable doubt about the accuracy of the examining physicians' diagnoses. For example, he explicitly discredited Dr. McNeer's conclusion that Miller suffered from such severe anxiety that he could not assume responsibility for a full-time job. "I was of the opinion that the body of [Dr. McNeer's] report suggested a lesser severity than he talked about in terms of the anxiety disorder." Record Transcript at 368. Dr. Weise concluded that, taken together, the claimant's records suggested that he "would be able to learn semi-skilled types of jobs." Id. at 372.

The ALJ characterized Dr. Weise's conclusions as a "comprehensive psychiatric record" resolving a "difference of opinion in this record." J.A. at 61. We are persuaded, however, that the ALJ erred by giving such obviously dispositive weight to this consulting psychiatrist's testimony. First, we think the witness' cold record evaluation of Miller's psychiatric condition was tainted by the ALJ's error in limiting the field of inquiry. Adhering to the ALJ's specific instructions, Dr. Weise did not consider the report prepared in January of 1987 at the Saint Albans center. The ALJ apparently instructed the witness to exclude the 1987 psychiatric evaluation on the basis of her earlier determination that Miller's eligibility for disability benefits had expired on September 30, 1986, and her related, implicit finding that the Saint Albans report did not constitute credible evidence of the claimant's condition prior to that date. As indicated, however, the Appeals Council later found that Miller last met the special earnings requirement on December 31, 1986. Thus, Miller was treated at Saint Albans only days after his insured status actually expired. The medical reports prepared shortly thereafter should therefore have been considered probative of the claimant's prior psychiatric condition, hence any preexisting disability. See, e.g., Stawls v. Califano, 596 F.2d 1209, 1213 (4th Cir.1979) (post-eligibility medical findings of psychiatric limitations may be probative of existence of disability in earlier, insured period). Dr. Weise's failure to take into account the Saint Albans findings therefore undermines the reliability (though not the honesty) of his testimony, a point most clearly confirmed by his own concession that consideration of the findings contained in the 1987 report might lead him to reevaluate his conclusion that Miller was not disabled. See Record Transcript at 374.

Second, and more importantly, the ALJ's exclusive reliance on Dr. Weise's testimony violates the settled rule in this circuit that "[t]he opinion of a non-examining physician cannot, by itself ..., serve as substantial evidence supporting a denial of disability." Leonard v. Schweiker, 724 F.2d 1076, 1078 (4th Cir.1983). See also Gordon v. Schweiker, 725 F.2d 231, 235 (4th Cir.1984) ("[T]he testimony of a non-examining, non-treating physician should be discounted and is not substantial evidence when totally contradicted by other evidence in the record."). Here, the non-treating, non-examining physician's testimony was obviously the dispositive evidence for the ALJ in rejecting Miller's claim of disability. We therefore find an absence of substantial evidence in the record to support the ALJ's rejection of the otherwise uncontradicted and consistent medical and psychological opinion that Miller's nonexertional impairments, either alone or certainly in combination with related exertional impairments, rendered him disabled to engage in substantial gainful employment.

Finding the evidence of disability uncontradicted, we conclude that on the record Miller was disabled and entitled to benefits within the covered period. Accordingly, we reverse the judgment of the district court and remand with instructions to remand to the Secretary for an award of proper benefits.

SO ORDERED.