In 30 depressed patients who had responded to total sleep deprivation therapy, morning naps led more frequently to relapses into depression than did afternoon naps. Longer naps were less detrimental than shorter ones, and there was no significant relationship between the effect of a nap on mood and its content of slow-wave-sleep. The amount of the rapid eye-movement sleep, too, was unrelated to clinical nap effects. Thus, some of the current theories on the relationship between sleep and depressive symptomatology are not supported by the data. Our results demonstrate the importance of nap timing, suggesting a circadian variation of propensity to relapse into depression.