We compared bolus injection of morphine 51 +/- 9 micrograms.kg-1 (M-S group) with 48 h-continuous infusion of bupivacaine and fentanyl mixture (BF-C group) for postoperative epidural analgesia in 100 patients who had undergone upper abdominal laparotomy. The epidural analgesia was started about 60 min before completion of surgery. The mixed solution, consisting of bupivacaine 48 ml (240 mg), fentanyl 24 ml (1.2 mg) and 0.9% saline (24 ml), was administered at a rate of 2 ml.h-1 by using a 100 ml balloon infuser. Analgesic effects (determined by visual analogue scale, verbal pain scores at rest and with movement, additional requests of other analgesics), comfort and side effects of the epidural analgesia were examined. The evaluation based on postoperative analgesia and comfortable feeling rated higher in the BF-C group compared with the M-S group (P < 0.01 to of 0.05). There were fewer requests for other postoperative analgesics in the BF-C group (P < 0.01 to 0.05). There was no significant difference in side effects between the two groups. The plasma fentanyl concentration in the BF-C group (n = 5) was maintained almost constant for the period from 24 to 49 hours after the start of infusion and was approximately 1.6 ng.ml-1. We conclude that continuous infusion of bupivacaine (2.5 mg.ml-1) and fentanyl (12.5 micrograms.ml-1) at a rate of 2 ml.h-1 is superior to bolus injection of morphine for postoperative analgesia.