Characteristic EEG sleep changes in depression are highlighted by a sleep continuity disturbance, delta sleep reduction, and a shortened REM latency. Since these findings have been derived primarily from only a few baseline recordings, questions regarding their persistence and/or variability have not been previously addressed. As part of an extensive set of investigations of EEG sleep in depression, we examined nightly the sleep of 12 hospitalized, non-delusional, primary depressives who were involved in a program of active psychosocial treatment intervention and received only placebo during a 5-week study period. EEG sleep findings revealed a relative lack of change across time, particularly in those parameters reported to be associated with a primary or 'biologic' depressive episode. While some degree of clinical improvement was noted, the group failed to achieve a state of remission or even partial remission as determined by the Hamilton Rating Scale. It appears that the major sleep alterations associated with such disorders persist for up to at least 5 weeks in the absence of pharmacologic or other somatic intervention.