Total sleep deprivation (TSD) for one whole night improves depressive symptoms in 40-60% of treatments. The degree of clinical change spans a continuum from complete remission to worsening (in 2-7%). Other side effects are sleepiness and (hypo-) mania. Sleep deprivation (SD) response shows up in the SD night or on the following day. Ten to 15% of patients respond after recovery sleep only. After recovery sleep 50-80% of day 1 responders suffer a complete or partial relapse; but improvement can last for weeks. Sleep seems to lead to relapse although this is not necessarily the case. Treatment effects may be stabilised by antidepressant drugs, lithium, shifting of sleep time or light therapy. The best predictor of a therapeutic effect is a large variability of mood. Current opinion is that partial sleep deprivation (PSD) in the second half of the night is equally effective as TSD. There are, however, indications that TSD is superior. Early PSD (i.e. sleeping between 3:00 and 6:00) has the same effect as late PSD given equal sleep duration. New data cast doubt on the time-honoured conviction that REM sleep deprivation is more effective than non-REM SD. Both may work by reducing total sleep time. SD is an unspecific therapy. The main indication is the depressive syndrome. Some studies show positive effects in Parkinson's disease. It is still unknown how sleep deprivation works.