In a prospective randomized multi-centre study, the mortality following internal fixation surgery for fracture of the upper femur was investigated in 538 elderly patients allocated to receive subarachnoid blockade or general (narcotic-relaxant) anaesthesia. The 28-day mortality was 6.6% with subarachnoid, and 5.9% with general, anaesthesia. The difference was not significant (95% confidence limits: -3.5 to +4.8). At 1 year following surgery, the mortality was 20.4%. Increasing age, ischaemic heart disease, cardiac failure, preoperative arrhythmias and poor ASA status were all associated with increases in early and long term mortality. A delay to surgery of more than 24 h from admission was also associated with an increased 28-day mortality. Senile dementia and admission other than from the patient's own home, were factors associated with a poorer long term outcome. From the point of view of mortality, subarachnoid anaesthesia did not appear to confer any advantages over general anaesthesia in non-prosthetic surgery for hip fracture in the elderly.