The outcome of treatment of 400 consecutive patients with ruptured intracranial aneurysms was assessed at 1 year. The patients were treated by a single surgeon over a period of 13 years. Data sheets completed as each patient was treated included a contemporary analysis of the reasons for any unsatisfactory outcome. Surgery was usually delayed for over 10 days from the last haemorrhage. Over the four successive 100-patient cohorts, in which the composition of the patient population remained unaltered, 1 year overall management mortality fell steadily from 38 to 24%. One year surgical mortality fell from 19 to 3%. The population of those operated on who were in Glasgow Outcome Score 5 at 1 year rose from 73 to 90% (from 51 to 71% for all patients). Of the 123 deaths, 89 occurred prior to operation, 24 after it. Thirty-five patients died from rebleeding prior to operation, but only eight of these occurred in patients judged fit for surgery at the time. All but one of the postoperative deaths resulted from technical problems related to the surgery. Over the successive cohorts, several factors indicated an improvement in operative efficiency, notably a fall in the proportion of cases with technical problems from 15 to 1%. We have demonstrated a steady improvement in management results, resulting largely from increasing operative experience. We do not believe that changes in overall management strategy, such as early surgery, would have any effect on overall outcome.