In the last decade all night EEG sleep research has attempted to delineate the sleep features most characteristics of depressive states occurring in adult life. We have reported that EEG sleep variables could be used to verify a diagnostic classification for affective syndromes. This EEG sleep generated schema significantly dichotimized each major clinical category using only two or three EEG sleep measures. REM latency and REM density were sufficient to separate out primary from secondary depressed patients. Sleep efficiency, REM sleep percent, and delta sleep percent discriminated between the psychotic and nonpsychotic subgroups in the primary depressive group. Furthermore, EEG sleep variables separated secondary depression with concurrent medical disease from secondary depression without medical disease using REM activity and intermittent nocturnal awakening as the requisite variables of discrimination. Other ongoing investigations have established that the REM intensity of the first REM period is increased in primary depression. The search for EEG sleep correlates of clinical response in depressed patients receiving tricyclic antidepressants has suggested that, the more rapid the suppression of REM sleep, the more likely that the patient will respond to tricyclic antidepressants. Finally, clinical response appears to be associated with a period of sustained elevation of REM latency.