We reviewed the literature on sleep in psychiatric disorders and evaluated the data by meta-analysis, a statistical method designed to combine data from different studies. A total of 177 studies with data from 7151 patients and controls were reviewed. Most psychiatric groups showed significantly reduced sleep efficiency and total sleep time, accounted for by decrements in non-rapid eye movement sleep. Rapid eye movement sleep time was relatively preserved in all groups, and percentage of rapid eye movement sleep was increased in affective disorders. Reduction in rapid eye movement sleep latency was seen in affective disorders but occurred in other categories as well. Although no single sleep variable appeared to have absolute specificity for any particular psychiatric disorder, patterns of sleep disturbances associated with categories of psychiatric illnesses were observed. Overall, findings for patients with affective disorders differed most frequently and significantly from those for normal controls.