Depression is a common source of distress in the elderly. Screening for depression allows for accurate diagnosis and treatment by clinicians and enables prevalence estimates to be used for monitoring morbidity and health services. A screening instrument is required that is both easily administered and has been validated among the heterogenous population of community-dwelling elderly in Israel. This study assesses the suitability of a short screening test with high face validity, the Short Geriatric Depression Scale (S-GDS), in a Jerusalem community sample (n = 285). The test yielded a 34% prevalence rate for depression, which is similar to rates found in community studies elsewhere. The screening test correctly classified 72% (95% confidence interval, 60 to 84%) of those with depression in a diagnosed subsample of 71 subjects. The specificity was only 57% (95% confidence interval, 44 to 70%) which was probably due to confounding with early dementia. The S-GDS was more likely to classify as depressed those with no formal education, those of Middle Eastern origin, and women. Higher levels of sensitivity and specificity can be obtained by calibrating, the cutoff score based on the level of education. The internal consistency of the test was adequate for the community sample as a whole, as well as among various demographic subgroups. The stability of test responses was also significant. Guidelines for the development of an instrument more appropriate for the Israeli population are suggested.