Studies on the behavioral consequences of rapid eye movement (REM) sleep deprivation in animals and humans are critically reviewed. In animals, converging evidence--some reasonably well controlled--indicates that REM sleep deprivation probably heightens central neural excitability and increased motivational behavior, but has nuclear or inconclusive effects on learning. In humans, evidence indicates that REM sleep deprivation is not dream deprivation and is not harmful to schizophrenic, depressed, or healthy subjects. Controversy continues about whether or not (some) schizophrenic patients respond abnormally to REM sleep deprivation by having no REM rebound. Controlled but unconfirmed work indicates that that endogenous, but not reactive, depressive patients are improved by REM sleep deprivation, a finding consistent with the animal behavioral consequences of the procedure and with the unique REM-depriving properties of efficacious antidepressant drugs.