The ventilatory response to CO2 was measured to evaluate the degree of respiratory depression after epidural sufentanil. After cesarean section performed with bupivacaine epidural anesthesia, 14 patients received either 30 micrograms (n = 7) or 50 micrograms (n = 7) of epidural sufentanil. Respiratory measurements were made before and 15, 45, and 120 min after sufentanil injection. The presence and severity of sedation and other nonrespiratory side effects were evaluated throughout the study. Plasma sufentanil assays were performed on blood samples obtained at frequent intervals during the first 2 h. Although changes in resting ventilation did not occur, both sufentanil doses depressed the ventilatory response to CO2. After sufentanil 30 micrograms, the slope of the CO2 response curve decreased significantly at 45 and 120 min (control value, 2.33 +/- 0.3 L.min-1.mm Hg-1 [mean +/- SEM] vs 1.61 +/- 0.24 and 1.72 +/- 0.15, respectively, P less than 0.05). After sufentanil 50 micrograms, significant decreases occurred at 15 and 45 min (control value, 2.84 +/- 0.71 vs 1.81 +/- 0.48 and 1.48 +/- 0.31 L.min-1.mm Hg-1, respectively). The mean maximal decrease in the slope occurred at 45 min and was more pronounced after 50 micrograms (-42.3% +/- 7.4%) than after 30 micrograms (-27.4% +/- 9.9%). Analgesia was similar in both groups. Side effects, particularly sedation, were more severe with the 50-micrograms dose. We conclude that 30 micrograms of epidural sufentanil is preferable to the higher dose with regard to both respiratory and nonrespiratory side effects. Even with the lower dose, monitoring of ventilation is advisable for a minimum of 2 h.