We have previously demonstrated that continuous epidural infusions of fentanyl without local anesthetics elicit analgesia by a systemic mechanism. In this study, we examined the hypothesis that, in the presence of epidural bupivacaine, continuous infusions of epidural fentanyl elicit analgesia by a spinal mechanism. Forty-eight nulliparous women in active labor participated in this prospective, randomized, double-blinded study. Women received lumbar epidural analgesia with 20–30 mL bupivacaine 0.125% until pain free. Subjects were then randomized to either IV or epidural (EPI) fentanyl infusion groups. Each infusion delivered fentanyl 30 μg/h. All women received an epidural infusion of bupivacaine at a rate of 20 mL/h, the concentration of which was determined by the response of the previous woman in the same group to the analgesic regimen used. Unlike previous studies that assessed the minimum local analgesic concentration (MLAC) for bolus administration at the initiation of analgesia, this study assessed MLACinfusion for the maintenance of analgesia throughout the first stage of labor. MLACinfusion was determined using the up-down sequential analysis described by Dixon and Massey. The MLACinfusion of epidural bupivacaine was 0.063% (95% confidence interval, 0.058–0.068) and 0.019% (95% confidence interval, 0.000–0.038) in the IV and EPI groups respectively. A continuous infusion of fentanyl was more than three times as potent when administered by the epidural than by the IV route. This marked increase in potency for the epidural route is highly suggestive for a predominantly spinal mechanism of action for infused epidural fentanyl under the conditions of this study.