Total sleep deprivation (TSD) exerts beneficial but only transient effects on mood in patients with a major depressive disorder (MDD). Though approximately 50 to 70% of depressed patients improve after sleep deprivation, the majority relapse after recovery sleep, some even after a short nap. One theoretical model postulates a critical period in the early morning hours where sleep is likely to induce a relapse, and nap studies indicate that sleep may be particularly 'depressogenic' at this time of day. A second model attributes the relapse to the release of non-REM sleep. We therefore compared the impact of an advanced sleep period (17:00-24:00 h) to a normal sleep period (23:00-06:00 h) on mood in patients who had responded to sleep deprivation. Less relapses into depression occurred after advanced sleep. Polysomnographic data showed that, as expected, normal sleep was characterized by a more pronounced improvement of sleep continuity and increased slow-wave sleep. The normal sleep group showed a stronger decrease in REM sleep density than the advanced sleep group compared with baseline. These data add to a growing body of evidence that the timing of sleep following successful sleep deprivation may be crucial for a stabilization of its antidepressant effect. Thus, avoidance of sleep during a "critical period' for more than a single night is necessary to provide a longer-lasting treatment modality.