The effect of hypobaric spinal anaesthesia or narcotic-halothane-relaxant general anaesthesia on the incidence of postoperative deep vein thrombosis was studied in 140 elective total hip replacements in a prospective randomised manner. Deep vein thrombosis was diagnosed using impedance plethysmography and the 125I fibrinogen uptake test, combined, in selected cases, with ascending contrast venography. The overall incidence of deep vein thrombosis was 20%. Nine patients (13%) developed deep vein thrombosis in the spinal group and nineteen (27%) in the general anaesthetic group (p less than 0.05). The incidences of proximal thrombosis and of bilateral thrombi were also less with spinal anaesthesia than with general anaesthesia. It is concluded that spinal anaesthesia reduces the risks of postoperative thromboembolism in hip replacement surgery. The presence of varicose veins, being a non-smoker and having a low body mass index were associated with an increased incidence of deep vein thrombosis.