In summary, the classical sleep disorders of nocturnal enuresis, somnambulism, the nightmare, and the sleep terror occur preferentially during arousal from slow-wave sleep and are virtually never associated with the rapid-eye-movement dreaming state. Original data are reported here which indicate that physiological differences from normal subjects, of a type predisposing the individual to a particular attack pattern, are present throughout the night. The episode, at least in the case of enuresis, appears to be simply a reinforcement of these differences to a clinically overt level. A number of features are common to all four sleep disorders. These had been shown previously to be attributable to the arousal itself. New data obtained by means of evoked potential techniques suggest that these common symptoms of the confusional period that follows non-REM sleep are related to alterations of cerebral reactivity, at least of the visual system. The symptoms which distinguish the individual attack types (that is, micturition, prolonged confusional fugues, overt terror) appear to be based upon physiological changes present throughout sleep which are markedly accentuated during arousal from slow-wave sleep. These changes may in some way be related to diurnal psychic conflicts. But, to date, it has proved impossible to demonstrate potentially causal psychological activity, dreaming or other forms of mental activity, or even a psychological void in sleep just preceding the attacks. The presence of all-night or even daytime predisposing physiological changes and the difficulty in obtaining any solid evidence of a preceding psychological cause explain, no doubt, why the results of efforts to cure the disorders at the moment of their occurrence (for example, by conditioning procedures in nocturnal enuresis) have been far from satisfactory. I stress the points that the attacks are best considered disorders of arousal and that the slow-wave sleep arousal episode which sets the stage for these attacks is a normal cyclic event. Indeed it is the most intense recurrent arousal that an individual regularly experiences. The most fruitful possibilities for future research would appear to be more detailed studies of those physiological changes that predispose individuals to certain types of attacks when they undergo intense arousal or stress; the reversal of these changes by psychological or pharmacological means; and more refined investigations of the physiological and psychological characteristics of the process of cyclic arousal from non-REM sleep.