This study examined the efficacy of patient-controlled epidural analgesia (PCEA) during labor and compared the suitability of three different PCEA solutions. After establishing effective epidural analgesia with 12 ml of 0.25% bupivacaine, 72 parturients in active labor were randomly assigned to one of four groups: physician-controlled continuous epidural infusion using 0.125% bupivacaine (CEI); PCEA using 0.125% bupivacaine (B); PCEA using 0.125% bupivacaine with fentanyl 1 micrograms/ml (BF); and PCEA using 0.125% bupivacaine with fentanyl 1 micrograms/ml and 1:400,000 epinephrine (BFE). The CEI infusion was begun at 12-16 ml/h and adjusted to maintain a T10 sensory level and adequate pain relief. PCEA pumps were programmed to deliver a 6 ml/h basal infusion, 4 ml on-demand boluses, 10-min lockout intervals between doses, and a 20 ml hourly limit. Hemodynamic parameters, sensory level, quality of analgesia, duration of labor, overall satisfaction, and Apgar scores did not differ among groups. Compared with CEI, PCEA with plain bupivacaine did not decrease total local anesthetic usage or average hourly infusion rates during labor. However, addition of fentanyl (groups BF and BFE) decreased hourly infusion requirements. Average hourly infusion rates were 13.0 +/- 1.1 ml/h (B), 10.6 +/- 0.6 ml/h (BF), and 9.6 +/- 0.5 ml/h (BFE); group B differs from others (P less than 0.05). No instance of respiratory depression or complication secondary to PCEA was observed. Mild pruritus occurred only with fentanyl-containing solutions, whereas dense motor block developed more frequently with the epinephrine-containing solution.(ABSTRACT TRUNCATED AT 250 WORDS)