OBJECTIVE We aimed to clarify whether the short-term adverse neonatal outcomes associated with epidural analgesia are due to the epidural analgesia itself or to the instrumental delivery. METHODS A retrospective case-control study was conducted to evaluate the relationship between epidural analgesia, labor length, and perinatal outcomes. A total of 350 pregnant women at term who delivered under epidural analgesia (cases) were compared with 1400 patients without epidural analgesia (controls). RESULTS Vacuum extraction (6.5 vs. 2.9 %) and cesarean section (19.9 vs. 11.1 %) were more frequently performed in the cases than controls (p < 0.001). Using a Kaplan-Meier algorithm, it was determined that the mean lengths of the 1st and 2nd stages of labor and the overall durations of labor and delivery were significantly longer in cases compared with controls. A Cox regression analysis showed that the longer labor remained even after adjustment for parity. The neonatal variables stratified by mode of delivery were not different in cases and controls, except for a slightly lower umbilical arterial pH in spontaneous delivery for the cases group. However, the Apgar scores and umbilical arterial pH were significantly lower in the neonates delivered by vacuum extraction compared with those in the neonates delivered by spontaneous delivery or cesarean section, regardless of whether epidural analgesia was performed. A multivariable analysis showed that vacuum extraction much more consistently affected the arterial pH than the analgesia itself (the β coefficients were -0.036 for epidural analgesia vs. -0.050 for vacuum extraction). CONCLUSIONS Epidural analgesia was associated with slowly progressing labor, thus resulting in an increased rate of instrumental delivery. This instrumental delivery appears to adversely affect the neonatal outcomes more strongly than the analgesia itself.