The relationship between insomnia and depression cannot be summarized as a symptom/disease relationship. It is well admitted now that sleep deprivation has an antidepressant effect on depressed patients. The effect is immediate, global but transient : the relapse occurs after subsequent sleep, diurnal nap or recovery night. On an other hand, sleep architecture is impaired in depressed patients, some of these alterations, especially in REM sleep, might have been considered specific of depressive disease. Antidepressant drugs exert an effect on sleep architecture which is different. This effect varies over time and generally tend to correct sleep impairment. This has led some authors to propose the hypothesis that sleep himself might be involved in the causal process of depression. Three main hypotheses will be considered, excluding those involving circadian rhythm impairment, according to their strong and weak points. For G. Vogel, an excess of REM sleep is the causal process in depression. As a matter of fact, he did show that selective REM sleep deprivation exerts an antidepressant effect following the same temporal profile as antidepressant drugs. An other argument is that the shortening of REM latency and the increased amount of REM sleep in the first half of the night are evidences of the excess of REM sleep and at last, most antidepressant drugs are REM suppressors and may act through REM sleep suppression. The second hypothesis is the process S deficiency proposed by Alexander Borbely, according to his two process model of sleep homeostasis. The impairment of process S, which is reflected by slow wave activity, is responsible for depression. This explain disruptions and shortening of sleep in depressed patients, REM sleep abnormalities being only secondary to the slow wave sleep reduction. More recently, D. Beersma and R. van den Hoofdakker proposed that non REM sleep might be depressogenic after an experiment of selective REM sleep deprivation in normals which showed that non REM sleep deprivation was also largely reduced. REM suppression effects might therefore also be attributed to non REM suppression. All these hypothesis must explain the effect of antidepressant drugs on sleep. There is a large heterogeneity of effects on slow wave, non REM and even REM sleep, hardly compatible with a causal role of sleep, REM or non REM.