The study of electroencephalogram (EEG) sleep in depressed patients before and after treatment with psychotherapy can distinguish episode-related and persistent biological features. With longitudinal follow-up, we can also assess whether EEG sleep measures are associated with recurrence of depression. In the current study, we examined EEG sleep during the depressed state and during symptomatic remission after treatment with interpersonal psychotherapy in 42 outpatients with major depression. Analyses included both visually-scored and computer-analyzed measures. Patients showed significant increases in sleep latency (p = .01) and rapid eye movement (REM) latency (p = .04) from baseline to remission, as well as a decrease in REM sleep percent (p = .03). Total delta EEG counts decreased from baseline to remission (p = .03), specifically in the second nonrapid eye movement (NREM) period (p = .03); as a result, the relative distribution of delta activity shifted toward sleep onset (i.e., increased delta sleep ratio; p = .03). Automated REM counts also decreased from depression to remission (p = .006). Compared to patients who remained well through one year of follow-up, those who suffered a recurrence of depression had less delta EEG activity at baseline and remission (p = .01), particularly in the lowest delta frequency band of 0.5-1.0 Hz. Specific components of sleep (total delta activity, delta ratio, REM activity) constitute episode-related biological features. Other components (slowest delta activity) may represent vulnerability factors for recurrence.