A retrospective study assessing the effect of epidural analgesia in labor on the incidence of cesarean section was performed. The first 500 consecutive nulliparas meeting the following criteria were included in this study: term (37 weeks or longer) and singleton gestation, cephalic presentation, spontaneous onset of labor, and 5 cm or less of cervical dilation on admission. Patients were grouped according to their rate of cervical dilation in early labor (greater than or equal to 1 cm/hr, and less than 1 cm/hr) and the timing of epidural placement (none, early, or late). There was no effect of epidural analgesia on the incidence of cesarean section for fetal distress. The incidence of cesarean section for dystocia was significantly greater (p greater than 0.000001) in the epidural group (15.6%) than in the nonepidural group (2.4%). The greatest effect of epidural analgesia on the incidence of cesarean section for dystocia was observed in nulliparas who dilated at slower rates (less than 1 cm/hr) in early labor and who had epidural analgesia placed at 5 cm or less of cervical dilation (20.6% versus 3.4%, relative risk of 6, p less than 0.0005). The increase of cesarean section for dystocia associated with epidural analgesia could not be accounted for when other possibly confounding variables were studied. Both the dilation rate prior to epidural placement and the cervical dilation at epidural placement were significantly correlated to frequency of cesarean section for dystocia (p less than 0.01). This study suggests that epidural analgesia in labor may increase the incidence of cesarean section for dystocia in nulliparas.