In a sample of 12 major depressive inpatients, endogenous subtype (8 primary and 4 secondary) defined by Research Diagnostic Criteria, we compared the sensitivity of four potential biological markers: latency of rapid eye movement (REM) sleep (recorded during at least 4 consecutive nights), dexamethasone suppression, and the clonidine and apomorphine tests. Shortened REM latency (less than 50 minutes during at least 1 night) identified 67% of depressives (87% of primary and 25% of secondary); nonsuppression after dexamethasone identified 50% of depressives (62% of primary and 25% of secondary); blunted growth hormone (GH) response after clonidine identified 75% of depressives (100% of primary and 25% of secondary); and blunted GH response after apomorphine identified 42% of depressives (62% of primary and 0% of secondary). Ninety-two percent of patients were correctly identified by at least one biological marker (100% of primary and 75% of secondary depressives). Of 67% of patients positive on at least two biological markers, all were primary depressives (100%). These four biological markers do not necessarily identify the same population, suggesting that their concurrent use may yield the highest level of diagnostic sensitivity.