Several abnormalities in sleep architecture have been described in schizophrenia. However, the question of whether sleep electroencephalographic (EEG) changes are influenced by treatment and phase of illness remains unclear. To examine the longitudinal stability of sleep data, we compared baseline sleep measures with measures obtained approximately 4 weeks and 1 year after the beginning of treatment in a series of schizophrenic patients. At the 4-week assessment, sleep continuity measures improved significantly; a modest increase in rapid eye movement (REM) latency was seen, but no other changes were found in sleep architecture. At the 1-year assessment, REM latency, REM time, and average automated REM counts increased. No significant changes were seen for slow-wave sleep (SWS) parameters at 1-year follow-up. These findings suggest that SWS parameters are relatively stable during follow-up, while REM parameters seem to change, perhaps in relation to phase of illness and treatment. SWS alterations may, at least in part, reflect more invariant, perhaps trait-related alterations in schizophrenia.