Notwithstanding recent advances in neuroimaging, EEG remains a major technique for investigation of the brain. Its main applications are in assessment of cerebral function rather than for detecting structural abnormalities. The principal clinical applications are in epilepsy, states of altered consciousness including postanoxic and traumatic coma, the parasomnias, dementias, toxic confusional states, cerebral infections, and various other encephalopathies. Abnormalities in EEG reflect general pathophysiological processes, raised intracranial pressure, cerebral anoxia, or oedema, epileptogenesis etc, and show little specificity for a particular disease. Consequently, they need to be interpreted in a particular clinical context; the use of routine EEG examination for screening purposes is rarely of value. Conversely, the investigation becomes most cost effective when applied to specific problems--for instance, monitoring serial changes in postanoxic coma or during open heart surgery, differential diagnosis (by telemetric ictal recordings) of epileptic and non-epileptic attacks, and providing early prediction of outcome after stroke. High technological standards and an individualised problem solving approach are prerequisites of a cost effective, reliable clinical EEG service. These are most likely to be achieved by a considered, selective referral policy, the use where necessary of prolonged complex procedures such as telemetry, and the avoidance of routine examinations of dubious clinical relevance.