A new screen for depression was compared with clinician diagnoses based on the Structured Clinical Interview for DSM-IV (SCID) as the standard. Post-menopausal women (n=436) completed the Burnam screen, a short version of the Center for Epidemiologic Studies Depression Scale (CES-D). The Burnam screen had a sensitivity of 74% and a specificity of 87% for current major depression and dysthymia, but the positive predictive value was low (20%) and the overall error rate was 14%. For lifetime mood disorders, sensitivity was very low for detecting affected subjects, even though specificity and positive predictive value were higher than for current conditions. Substituting a more sensitive cutpoint slightly improved the screen's ability to detect subjects with lifetime mood disorders. Even algorithms that used coefficients optimized for these data gave little improvement in the psychometric properties of the Burnam screen. These results re-emphasize the difficulty of using a one-stage screen to detect accurately a depressive diagnosis.