Twenty-six patients were subjected to temporal lobectomy for complex partial seizures resistant to anticonvulsant control. The criteria for selection were simple. There should be EEG evidence of epileptiform activity arising predominantly in one temporal lobe. It was desirable that psychometric tests should confirm the localisation by showing defects of function in this lobe, and necessary that they should demonstrate normal function in the other. There should be no radiological evidence of a gross structural lesion. Twenty-four patients have been followed for 5 to 17 years, two being lost at 3 and 4 years. After operation, 50% were completely free from seizures, and in a further 38% seizures were reduced in frequency by at least 75%. There was no mortality, and the only morbidity was a minor quadrantic hemianopia, usually unnoticed by the patient. A significant improvement in mental state was noted in many patients. It is concluded that to use simple criteria for selection of patients for temporal lobectomy gives worthwhile results. There are many patients in the community who could be selected for surgery on these criteria with a high probability that their condition would be improved.