In an investigation on deep venous thrombosis and pulmonary embolism, where neither dextran nor antithrombotic drug prophylaxis were employed, 30 patients undergoing total hip replacement were randomly allotted to one of two groups receiving either epidural or general anaesthesia. The epidural group (n = 15) was given 0.5% bupivacaine with epinephrine (5 micrograms/ml) and this was prolonged into the postoperative period for pain relief. The general anaesthesia group (n = 15) was operated on under artificial ventilation with nitrous oxide/oxygen via an endotracheal tube and intravenously administered fentanyl and pancuronium bromide. In this group of patients narcotic analgesics (ketobemidone) were given intramuscularly on demand for pain relief postoperatively. The frequency of deep venous thrombosis involving the femoral veins, as observed at phlebography, was significantly lower in patients receiving continuous epidural block (3 of 15; 20%), than in those receiving general anaesthesia and parenteral analgesics postoperatively (11 of 15; 73%). Further, the frequency of pulmonary embolism, as determined by pulmonary perfusion lung scanning, was lower in patients receiving continuous epidural block (2 of 15) than in the general anaesthesia group (7 of 15). Possible explanations for these findings are discussed, including a hyperkinetic lower limb blood flow and lower fibrinolysis inhibition activity in patients given epidural block. Lower blood transfusion requirements in patients given epidural block might also play a role, as well as a "stabilizing" effect of local anaesthetics on platelets, leukocytes and endothelial cells.