The assumption that sleep dysregulation is more than a mere epiphenomenon of depression is based on several observations: sleep disturbances are strongly associated with the depressive state; a number of sleep manipulations can alleviate symptoms of depression in some patients; and the majority of antidepressants bring about remarkable changes in sleep polygraphic variables. An obvious question is whether changes in sleep physiological processes are intimately involved in the pathogenesis and recovery from depression. One way to elucidate the link between sleep and depression is to examine whether the influence of antidepressants on sleep is related to clinical improvements in depressives. For that purpose, the effects of antidepressants on EEG sleep and their importance for the treatment of depression are summarized against the background of two existing hypotheses concerning the link between sleep and depression: one hypothesis concerning the role of REM; the other concerning the role of non-REM sleep. EEG sleep studies on the use of antidepressants in depressives have not produced clear evidence of the involvement of REM sleep or non-REM sleep in the mechanisms underlying clinical change. Furthermore, the role of sleep physiological mechanisms during treatment with antidepressants is still unclear. To interpret the effects of antidepressants on EEG sleep in terms of sleep physiological processes more fundamental sleep research is necessary. Also, more comparative studies of antidepressants with similar therapeutic effects but different pharmacological profiles are needed in both healthy and depressed subjects to further quantify the impact of EEG sleep modification in the recovery from depression and to differentiate between pharmacological and sleep-related aspects.