In recent years the effectiveness of hypnotic drugs has had to be assessed in terms of a greatly increased knowledge of the physiology and pathology of sleep. The normal pattern of sleep and wakefulness involves a cyclic alternation between three rather than two basically dissimilar states of the brain and body - alert wakefulness, rapid-eye-movement (REM) sleep and non-rapid-eye-movement (NREM) sleep. The pattern of this alternation in individual people results from the interaction of many influences - biological (including genetic, early developmental and later degenerative influences), psychological, social and environmental factors, various physical and psychiatric disorders, and most drugs which affect the central nervous system. The quality of sleep is not related in any simple or constant manner either to its duration or to the proprotions of time spent in each stage of sleep. Among the disorders of sleep, insomnia is a far more common problem of medical management than are enuresis, narcolepsy, somnambulism or nightmares. With a few exceptions, most hypnotic drugs now in widespread use cease to be effective in treating insomnia after the first few nights. However, the ineffective treatment is often continued because insomnia will be even worse during the initial period of drug withdrawal. These factors and the toxicity of hypnotic drugs when taken in overdose make the long-term treatment of insomnia more difficult than was previously supposed. Barbiturates should no longer be prescribed. Some of the non-barbiturates, such as glutethimide and methaqualone, have no advantage over the barbiturates. The benzodiazepine hypnotics, nitrazepam and flurazepam, are less toxic in overdose and are relatively effective in treating insomnia. Chloral hydrate and its derivates are useful alternative drugs for short-term use. Measures to improve sleep without drugs deserve greater emphasis than they have had in the past.