The study assessed the effect of screening for mental disorder by means of the General Health Questionnaire (GHQ) on the rate of detection of mental disorder by fourteen physicians in a primary-care clinic. After completing the GHQ, patients were randomised into control (722 patients) and experimental (730 patients) groups. GHQ results of the experimental group were made available to the physicians; those of the control group were not. Sociodemographic factors influenced the physicians' rate of diagnosis of mental disorders (rates were lower for men, students, and patients with at least a partial college education than in subjects who had a low income, less than 7 years of school, or were widowed) but there was no difference between control and experimental groups (16.8% vs 16.0%). Among patients with a prior diagnosis of a mental disorder, twice as many were found to have mental disorders by the physicians as by the GHQ (70% vs 33%).