BACKGROUND: There are numerous instruments for screening for depression. A feasible screen is good at both recognising and predicting depression. AIMS: To study the ability of the Depression Scale and its items to recognise and predict a depressive episode. METHOD: A sample of patients attending primary care was examined in 1991-992 and again 7 years later. The accuracy of the Depression Scale at baseline and at follow-up was tested against the Short Form of the Composite International Diagnostic Interview (CIDI-SF) diagnosis of depression at follow-up. The sensitivity and specificity of the Depression Scale and its items were assessed. RESULTS: Both baseline and follow-up Depression Scale scores were consistent with the CIDI-SF diagnoses. It was possible to find single items efficient at both recognising and predicting depression. CONCLUSIONS: The Depression Scale is a useful screening instrument for depression, with both diagnostic and predictive validity.