This study aims to test a short form of the Center for Epidemiological Studies--Depression Scale (CES-D) which can be a useful screening tool for depressive symptomatology in epidemiological studies of elderly patients. The study was conducted on 2792 subjects from the PAQUID (Personnes Agées QUID?) cohort, an epidemiological survey of community dwellers living in South-West France. CES-D items with high sensitivity and good specificity were selected for the short form, then the best cut-off scores were determined with Receiver Operating Characteristics (ROC) curves. The external validity of the 5-item scale was then assessed against the full scale at different PAQUID follow-ups. Sensitivity was 99% and specificity 81% for detecting depressive symptomatology when compared to the 20-item scale. The external validity on the different follow-ups was good, yielding a sensitivity varying from 95 to 100%, and a specificity from 83 to 89%. In conclusion, the 5-item CES-D is a simple, rapid and reliable tool which could be useful for screening depressive symptoms in epidemiological studies of the elderly.