We describe demographic, clinical, laboratory and aetiological findings in 93 consecutive patients with rapid eye movement (REM) sleep behaviour disorder (RBD), which consists of excessive motor activity during dreaming in association with loss of skeletal muscle atonia of REM sleep. The patients were seen at the Mayo Sleep Disorders Center between January 1, 1991 and July 31, 1995. Eighty-one patients (87%) were male. The mean age of RBD onset was 60.9 years (range 36-84 years) and the mean age at presentation was 64.4 years (37-85 years). Thirty-two per cent of patients had injured themselves and 64% had assaulted their spouses. Subdural haematomas occurred in two patients. Dream content was altered and involved defence of the sleeper against attack in 87%. The frequency of nocturnal events decreased with time in seven untreated patients with neurodegenerative disease. MRI or CT head scans were performed in 56% of patients. Although four scans showed brainstem pathology, all of these patients had apparently unrelated neurodegenerative diseases known to be associated with RBD. Neurological disorders were present in 57% of patients; Parkinson's disease, dementia without parkinsonism and multiple system atrophy accounted for all but 14% of these. RBD developed before parkinsonism in 52% of the patients with Parkinson's disease. Five of the 14 patients with multiple system atrophy were female, and thus the strong male predominance in RBD is less evident in this condition. Psychiatric disorders, drug use or drug withdrawal were rarely causally related to RBD. Clonazepam treatment of RBD was completely or partially successful in 87% of the patients who used the drug. We conclude that RBD is a well-defined condition and that descriptions from different centres are fairly consistent. It is commonest in elderly males and may result in serious morbidity to patients and bed partners. There is a strong relationship to neurodegenerative disease, especially Parkinson's disease, multiple system atrophy and dementia, and neurologists should explore the possibility of RBD in patients with these conditions. RBD symptoms may be the first manifestations of these disorders and careful follow-up is needed. Neuroimaging is unlikely to reveal underlying disorders not suspected clinically. We confirm the effectiveness of clonazepam, but note that attention to the safety of the bed environment may be sufficient for patients with contraindications to the drug.