Manipulations of the sleep-wake cycle, whether of duration (total or partial sleep deprivation [SD]) or timing (partial SD, phase advance), have profound and rapid effects on depressed mood in 60% of all diagnostic subgroups of affective disorders. Relapse after recovery sleep is less when patients are receiving medication; it may be prevented by co-administration of lithium, pindolol, serotonergic antidepressants, bright light, or a subsequent phase advance procedure. Diurnal and day-to-day mood variability predict both short-term response to SD and long-term response to antidepressant drug treatment. These mood patterns can be understood in terms of a "two-process model of mood regulation" based on the model well established for sleep regulation: the interaction of circadian and homeostatic processes. The therapeutic effect of SD is postulated to be linked to changes in disturbed circadian- and sleep-wake-dependent phase relationships and concomitant increase of slow-wave-sleep pressure; additionally, SD-induced sleepiness may counteract the hyperarousal state in depression. This model has the advantage of providing a comprehensive theoretical framework and stringent protocols ("constant routine," "forced desynchrony") to dissect out specific disturbances. Many aspects tie in with current serotonergic receptor hypotheses of SD action. A treatment inducing euthymia in severely depressed patients within hours is an important therapeutic option that has come of age for clinical use.