In a prospective study of 50 patients subjected to major abdominal surgery, the frequencies of postoperative deep vein thrombosis and pulmonary embolism were analysed. The patients were randomized to one of two groups receiving either neurolept anaesthesia or neurolept anaesthesia combined with thoracic epidural analgesia. Five patients were excluded. No special anti-thrombotic prophylaxis was administered. Deep vein thrombosis was diagnosed with the 125I-fibrinogen test and pulmonary embolism with pre- and postoperative lung perfusion scintigraphy combined with lung X-ray. Patients with positive scintigraphy were subjected to pulmonary angiography for verification of the diagnosis. Deep vein thrombosis was treated when diagnosed. The frequency of deep vein thrombosis was equal in both groups (38%). No patient with pulmonary embolism was recorded during the first seven days after operation. It is concluded that the addition of thoracic epidural analgesia to neurolept anaesthesia does not alter the postoperative frequency of deep vein thrombosis in patients subjected to major abdominal surgery. Early diagnosis and treatment of postoperative deep vein thrombosis might prevent pulmonary embolism. Problems encountered in the diagnosis of postoperative pulmonary embolism are discussed.