The ability of two depression scales, the Center for Epidemiologic Studies Depression Scale (CES-D) and the Beck Depression Inventory (BDI), to identify cases of DSM-III-R major depression and dysthymia was investigated in a large, community sample of high school students. Receiver operating characteristics analyses indicated that different caseness criteria should be used for boys and girls for both the CES-D and the BDI. Internal consistency-reliability and sensitivity and specificity for detecting current episodes of current depression and dysthymia were adequate and comparable to those found with adult samples, but both the CES-D and the BDI generated many false positives. Multiple screening using the "serial" strategy increased positive predictive power substantially for both the CES-D and the BDI, whereas using the "parallel" strategy had very little effect on the efficacy of the two screeners. The results indicate that neither the BDI nor the CES-D should be used by themselves as methods for case ascertainment in either epidemiological or experimental studies, although the BDI does function somewhat better than the CES-D as a screener.