OBJECTIVE To assess the importance of non-rapid eye movement (NREM) sleep disturbance in major depressive disorder (MDD) patients using cyclic alternating pattern (CAP) analysis, and to determine the usefulness of CAP analysis in evaluating treatment effect. METHODS Baseline sleep-staging data and CAP analysis of NREM sleep was compared in 15 MDD patients (Hamilton depression scale score>20) and normal controls. Longitudinal evaluation of sleep changes using similar analysis during a treatment trial was also performed. METHODS A single-blinded researcher scored and analyzed the sleep of MDD and age-matched normal controls at baseline and during a treatment trial using the international scoring system as well as CAP analysis. RESULTS MDD patients had evidence of disturbed sleep with both analyses, but CAP analysis revealed more important changes in NREM sleep of MDD patients at baseline than did conventional sleep staging. There was a significant decrease in CAP rate, time, and cycle and disturbances of phase A subtype of CAP. NREM abnormalities, observed by CAP analysis, during the treatment trial paralleled subjective responses. Analysis of subtype A phase of CAP demonstrated better sleep improvement. CONCLUSIONS CAP analysis demonstrated the presence of more important NREM sleep disturbances in MDD patients than did conventional sleep staging, suggesting the involvement of slow wave sleep (SWS) in the sleep impairment of MDD patients. Improvement of NREM sleep paralleled subjective mood improvement and preceded REM sleep improvement. CAP analysis allowed objective investigation of the effect of treatment on sleep disturbances.