In a survey of 108 subjects with a history of epilepsy in a hospital for the mentally handicapped, administration of both phenobarbitone and phenytoin was associated with low serum calcium and plasma 25-hydroxyvitamin D (25-(OH)D) levels in female subjects only. Intake of phenytoin (as mg/kg body weight) in female subjects exceeded that in males by 22 per cent, whilst the intake of phenobarbitone was 37 per cent higher. The doses of phenobarbitone and phenytoin were each inversely related to plasma 25-(OH)D concentration, but anticonvulsant drug dosage did not correlate with the magnitude of the decline of plasma 25-(OH)D concentration in winter (November-February). No influence of sodium valproate was detected on serum calcium or on plasma 25-(OH)D levels. Limited exposure to ultraviolet irradiation (UVR) or oral administration of vitamin D restored plasma 25-(OH)D to normal levels and healed osteomalacia in a subject with tuberous sclerosis. In this subject, fit frequency declined in response to UVR and to a lesser extent in response to oral vitamin D, despite the attainment of similar levels of serum calcium and of plasma 25-(OH)D. Serum calcium levels in the other 108 subjects were lower in those experiencing the most frequent fits, but serum calcium could not be restored to levels found in subjects not receiving anticonvulsant drugs unless supraphysiological doses of vitamin D were given. Vitamin D deficiency in the epileptic population receiving drugs was assessed by the response of alkaline phosphatase to vitamin D administration. A consistent fall of serum alkaline phosphatase was found only if the initial level exceeded 175 per cent of the normal value established by reference to a population not receiving phenobarbitone or phenytoin. By this criterion five out of 45 subjects (11 per cent), aged nine to 36 years were vitamin D deficient.