In 1987, Yeager et al. reported that intraoperative epidural anesthesia with local anesthetics and postoperative epidural analgesia with opiates diminished postoperative morbidity. In our first clinical trial on this topic, the better postoperative analgesia with epidural bupivacaine-fentanyl failed to improve the outcome after major abdominal operations over that obtained with parenteral piritramide. This randomized controlled investigation was designed to assess whether intraoperative epidural anesthesia with bupivacaine plus light general anesthesia and postoperative epidural analgesia with morphine would diminish the overall rate of postoperative complications after major abdominal operations compared with general anesthesia (without epidural) followed by patient controlled analgesia with morphine, and with intraoperative epidural anesthesia with bupivacaine and light general anesthesia followed by postoperative bupivacaine-morphine analgesia. METHODS. A total of 292 patients undergoing infrarenal aortic bypass operation, gastric resection, gastrectomy, duodenum-preserving pancreatic resection, Whipple's operation or cystectomy and neobladder formation were randomly divided into three groups: 1. PCA group (patient controlled analgesia, n = 107): patients were operated on under general anesthesia (midazolam, fentanyl, N2O/O2, if necessary with addition of halothane, enflurane or isoflurane; muscle relaxation with pancuronium bromide). Postoperative management consisted in patient-controlled analgesia with morphine (Prominject), bolus 2 mg, lock-out 5 min (recovery room, intensive care unit) or 15 min (surgical ward). 2. EBM group (epidural bupivacaine+morphine, n = 95): operation under light general anesthesia (midazolam, low-dose fentanyl, N2O/O2, pancuronium bromide). In addition, a mixture of bupivacaine (0.25%) and morphine (60 micrograms/ml) was infused (approximately 0.1 ml/kg.h) via an epidural catheter during and after the operation (approximately 72 h). 3. EM group (epidural morphine, n = 90): operation under the same kind of general-epidural anesthesia as in the EBM group. Postoperatively, epidural injection of morphine (0.05 mg/kg in 10 ml of saline) on request up to the 3rd postoperative day. Quality of analgesia (at rest and when patients coughed vigorously), strength of cough, and rate-pressure product were recorded at 8:00 h, 12:00 noon, 16:00 h and 20:00 h on the 1st, 2nd and 3rd postoperative days. Incidence and intensity of all postoperative complications (cardiovascular, pulmonary, renal and other organ failure, reoperations, major infection, sepsis, thromboembolism, metabolic and mental disturbances) were assessed from the day of operation until discharge or death (n = 10), respectively. RESULTS AND DISCUSSION. In the PCA and EM groups analgesia was equal but of slightly inferior quality compared with the EBM group. The ability to cough was best in the EBM group and significantly worse in the PCA and EM groups, with no difference between the last two. (ABSTRACT TRUNCATED AT 400 WORDS)