Epidural blood flow was measured in seven patients undergoing elective abdominal surgery during combined lumbar epidural and general anesthesia. After an initial dose of 20 ml plain bupivacaine 0.5%, a continuous epidural infusion of bupivacaine 0.5% (8 ml/hr) was given for 16 hours for postoperative pain relief. The epidural blood flow was measured by a local 133Xe clearance technique in which 15-35 MBq 133Xe diluted in 1 ml saline was injected through the epidural catheter on the day before surgery (no bupivacaine), 30 minutes after the initial dose of bupivacaine on the morning before surgery, and 8, 12, and 16 hours later during the continuous infusion. Initial blood flow was 6.0 +/- 0.7 ml/min per 100 g tissue (mean +/- SEM). After epidural bupivacaine, blood flow increased in all seven patients to 7.4 +/- 0.7 ml (P less than 0.02). Initial level of sensory analgesia was T4.5 +/- 0.17 (mean +/- SEM). Postoperatively, two patients maintained the initial level of sensory analgesia and low pain score throughout the 16-hour study. In these two patients epidural blood flow remained constant after the initial increase. Flow increased further to 10.3 +/- 0.8 ml/min per 100 g tissue (P less than 0.03) in the other five patients as the level of sensory analgesia regressed postoperatively. These data suggest that changes in epidural blood flow during continuous epidural infusion of bupivacaine, and thus changes in rates of vascular absorption of bupivacaine from the epidural space, may be an important factor contributing to differences in rates of regression of sensory analgesia.