A hundred patients scheduled for elective abdominal surgery were randomized to either general anaesthesia (low-dose fentanyl) and systemic morphine for postoperative pain or combined general anaesthesia and epidural analgesia with etidocaine 1.5% intraoperatively (T4-S5) and bupivacaine 0.5% 5 ml/4 h for 24 h and morphine 4 mg/12 h for 72 h. Postoperative pain was better controlled by the epidural regimen (P less than 0.0001). We found no significant reduction in postoperative mortality (6% to 2%), pneumonia (28% to 20%), cardiac dysrhythmia (10% to 5%) and wound complications (14% to 11%) by the epidural analgesic regimen. The incidence of deep venous thrombosis (125I-fibrinogen scan) was 32% after general anaesthesia and low-dose heparin and 34% after epidural analgesia with no prophylactic antithrombotic treatment (P greater than 0.9). Postoperative weight loss and decrease in serum-albumin and serum-transferrin, as well as the reduction in haemoglobin and the need for postoperative transfusions, were similar in the two groups. Convalescence, as assessed by postoperative fatigue, restoration of bowel function (flatus, bowel movement and food intake) and the time until the patients were self-aided at their preoperative level, was not reduced by epidural analgesia. Since 50% of the patients in each group suffered from one or more of the above-mentioned postoperative complications, this epidural regimen was not effective in reducing postoperative morbidity after major abdominal surgery despite the achievement of adequate pain relief.