The prevalence of psychiatric disorder in 100 patients attending their general practitioner was found to be 33% based on the criteria of the DSM-3 diagnostic system. Using the DSM-3 diagnosis as a yardstick, the performance of the hospital anxiety depression scale was compared with that of the general health questionnaire. Relative operating coefficient analysis showed good discrimination between ;cases' and ;non-cases' for both questionnaires, and the optimum threshold score was found to be eight for the hospital anxiety depression scale and five for the general health questionnaire. Using these threshold scores the positive predictive value was 81% for the hospital anxiety depression scale and 77% for the general health questionnaire. The hospital anxiety depression scale appeared more sensitive than the general health questionnaire (90% versus 77%) and simpler to complete. In addition, it does not require a different threshold score for each population studied. The use of screening questionnaires in general practice is discussed.