Previous investigations have indicated that one of the most consistent EEG sleep findings in depressive patients has been a shortened REM latency. On the basis of these studies, we have concluded that with the exception of drug withdrawal states (such as CNS depressant or amphetamine withdrawal and narcolepsy) shortened REM latency points to a strong affective component in the patient's illness. Short REM latency has also been observed in patients suffering from schizo-affective illness as well as in certain schizophrenic patients who require tricyclic antidepressants in their management. Furthermore, this psychobiologic marker is a persistent, rather than a transient phenomenon, and can be observed over a period of several weeks unless a patient's condition becomes more favorable through clinical intervention. This present report indicates that short REM latency is found in virtually all primary depressive illness and is absent in secondary depression. Thus, REM latency appears to be a dependable, measurable marker for diagnosing primary depression, and we argue that the phenomenon is independent of age, drug effect and changes in other sleep parameters. It is expected that EEG sleep and motor measurements can yield further significant data and improve differential diagnosis in psychiatry, in much the same way that laboratory data support other medical specialities.