OBJECTIVE Continuous epidural analgesia may be considered in liver resection but is often avoided because of possible coagulopathies and the risk of epidural hematoma in the postoperative period. On the other hand, there is no coagulation defect during the surgery. Effective prevention of postoperative pain may require continuous sensory ablation throughout the surgery event. METHODS A prospective, randomized, double-blind study was conducted to evaluate the efficacy of intraoperative epidural anesthesia on postoperative morphine consumption via patient-controlled analgesia after liver surgery in 2 groups of patients. One group (epidural) received, intraoperatively, thoracic epidural bupivacaine perfusion (0.5% at 3 mL/hr) added to preoperative intrathecal morphine (0.5 mg) and fentanyl (15 microg). The other group (placebo) was administered the same intrathecal narcotics but with a sham epidural. Forty-four patients scheduled for major liver resection (> or =2 segments) were recruited. Patient-controlled analgesia morphine consumption, pain at rest and with movement, sedation, nausea, pruritus, and respiratory frequency were evaluated at 6, 9, 12, 18, 24, 36, and 48 hrs after intrathecal morphine injection. RESULTS Patients in the placebo group consumed twice as much morphine during each time interval than patients in the epidural group (at 48 hrs: 123 [SD, 46] vs 59 [SD, 25] mg; P < 0.0001). Pain evaluation on visual analog scale at rest and on movement was lower in the epidural group (P = 0.017 and P = 0.037). CONCLUSIONS Intraoperative thoracic epidural infusion of bupivacaine, added to intrathecal morphine, decreased postoperative morphine consumption with better pain relief compared with the placebo.