To compare the efficacy of patient-controlled lumbar and thoracic epidural sufentanil, 22 patients scheduled for elective thoracotomy were assigned randomly to receive sufentanil via either a lumbar or a thoracic epidural catheter. For 24 h postoperatively, the patients received analgesia only by patient-controlled epidural sufentanil. There were no significant differences in the visual analog scale (VAS) for pain between the two groups at 8 and 24 h postoperatively. Nausea and pruritus were minimum, requiring treatment less than once per 24-h period in either group. The forced vital capacity (FVC) measured at 24 h (as a percentage of baseline FVC) showed no significant difference between the lumbar and thoracic groups (44.7 ± 3.8 and 41.7 ± 5.5; P = 0.68). The total sufentanil used by the lumbar and thoracic groups was not significantly different (196 ± 25.2 μg and 157 ± 28.6 μg; P = 0.32). We conclude that there is no clinical advantage of thoracic over lumbar epidural sufentanil in the thoracotomy patient with respect to quality of analgesia, amount of sufentanil used, severity of side effects, or postoperative pulmonary function.