In studies comparing the performance of psychometric instruments and general practitioners in the identification of psychological disorders, authors usually treat the psychometric instrument as the gold standard. Some patients may have no psychiatric diagnosis and normal scores on self-report measures of distress, but still benefit from detection and treatment of their psychosocial problems. However, physicians may be spending valuable time identifying problems in patients who have no disability. The extent and implications of the discrepancy between clinician assessment and standard instruments requires further exploration. Adult patients of 40 family physicians completed the General Health Questionnaire (GHQ-28) before their visit. Immediately following the visit, physicians, who were blind to the patient's GHQ score, indicated whether they had detected any signs or symptoms of anxiety, depression, somatization, or other psychosocial problems. Of the 1,011 primary care patients that participated, 439 had normal GHQ-28 scores. Physicians detected psychological problems in 177 (38.3%) of the 439. In bivariate analyses, poorer general and mental health (as measured by SF-36) was associated with higher detection rates. The patient's belief that there was a psychological component of his or her problem (OR = 2.50), being in a marital relationship (OR = 1.87), and the physician's perception of the seriousness of the problem (OR = 1.84) were associated with detection. Detection was less frequent when the physician did not know the patient well (OR = 0.69), and when the physician was a woman (OR = 0.46). For the 28% of patients who themselves perceived a psychological element of their problem, physician detection was probably appropriate. However, it is unlikely that detection of the remaining patients was beneficial to the patients.