Electroencephalographic (EEG) sleep measures have been examined as predictors of therapeutic response in patients with major depression. Although some studies have reported that EEG sleep measures are predictive of a favorable outcome with medications, two recent studies found no differences in the baseline sleep characteristics of responders and nonresponders to psychotherapy. To clarify this issue, we compared baseline EEG sleep in a group of patients with recurrent depression who responded to interpersonal psychotherapy (n = 19) and a comparable group who did not respond (n = 18). Baseline ratings of depression severity did not differ in the groups, but some differences in baseline sleep were noted. Psychotherapy nonresponders had longer sleep latencies, lower sleep efficiency, and increased automated measures of phasic rapid eye movement (REM) activity. In addition, the two groups had different EEG sleep adaptation patterns for REM latency and phasic REM density measures across the two study nights. These preliminary results suggest that baseline EEG sleep patterns, as well as the pattern of laboratory adaptation, may differ for depressed patients who respond to psychotherapy and those who do not.