OBJECTIVE To determine the efficacy of oral nimodipine in reducing cerebral infarction and poor outcomes (death and severe disability) after subarachnoid haemorrhage. DESIGN Double blind, placebo controlled, randomised trial with three months of follow up and intention to treat analysis. To have an 80% chance with a significance level of 0.05 of detecting a 50% reduction in an incidence of cerebral infarction of 15% a minimum of 540 patients was required. SETTING Four regional neurosurgical units in the United Kingdom. PATIENTS In all 554 patients were recruited between June 1985 and September 1987 out of a population of 1115 patients admitted with subarachnoid haemorrhage proved by the results of lumbar puncture or computed tomography, or both. The main exclusion criterion was admission to the neurosurgical units more than 96 hours after subarachnoid haemorrhage. There were four breaks of code and no exclusions after entry. One patient was withdrawn and in 130 treatment was discontinued early. All patients were followed up for three months and were included in the analysis, except the patient who had been withdrawn. INTERVENTIONS Placebo or nimodipine 60 mg was given orally every four hours for 21 days to 276 and 278 patients, respectively. Treatment was started within 96 hours after subarachnoid haemorrhage. END POINTS Incidence of cerebral infarction and ischaemic neurological deficits and outcome three months after entry. MEASUREMENTS Demographic and clinical data, including age, sex, history of hypertension and subarachnoid haemorrhage, severity of haemorrhage according to an adaptation of the Glasgow coma scale, number and site of aneurysms on angiography, and initial findings on computed tomography were measured at entry. Deterioration, defined as development of a focal sign or fall of more than one point on the Glasgow coma scale for more than six hours, was investigated by using clinical criteria and by computed tomography, by lumbar puncture, or at necropsy when appropriate. All episodes of deterioration and all patients with a three month outcome other than a good recovery were assessed by a review committee. MAIN RESULTS Demographic and clinical data at entry were similar in the two groups. In patients given nimodipine the incidence of cerebral infarction was 22% (61/278) compared with 33% (92/276) in those given placebo, a significant reduction of 34% (95% confidence interval 13 to 50%). Poor outcomes were also significantly reduced by 40% (95% confidence interval 20 to 55%) with nimodipine (20% (55/278) in patients given nimodipine v 33% (91/278) in those given placebo). CONCLUSIONS Oral nimodipine 60 mg four hourly is well tolerated and reduces cerebral infarction snd improves outcome after subarachnoid haemorrhage.