This study was designed to evaluate the potential fentanyl-sparing effect of a dilute local anesthetic, bupivacaine, administered in fixed combinations with fentanyl for post-thoracotomy analgesia via a continuous thoracic epidural infusion. Forty adult patients scheduled for thoracotomy were randomly allocated in a double-blind fashion to receive an epidural infusion containing 0, 0.03, 0.06, or 0.125% bupivacaine in combination with fentanyl (4 micrograms/mL). The epidural infusions were initiated in the operating room at 10 mL/hr. During the first 24 hours, there were no between-group differences in pain scores. Total fentanyl use was significantly decreased 24% to 33% in all bupivacaine treatment groups. However, fentanyl plasma levels at 24 hours were not significantly different between groups. Arterial blood gas measurements performed on the morning after surgery revealed significant reductions in PaCO2 values, 38 +/- 4, 36 +/- 4, 37 +/- 4 mmHg for 0.03, 0.06, and 0.125% bupivacaine groups respectively, versus 44 +/- 6 for the plain fentanyl group. Arterial pH values were significantly higher in all bupivacaine treatment groups. These findings suggest that the combination of dilute bupivacaine with fentanyl for thoracic epidural analgesia for post-thoracotomy pain may have beneficial effects on pulmonary gas exchange.