A double-blind, placebo-controlled study was designed to compare the efficacy of demand-dose patient-controlled epidural analgesia (PCEA) with continuous epidural infusion (CEI) for treatment of pain during labor and delivery. Forty patients were randomized to receive 0.125% bupivacaine with fentanyl (2 micrograms/mL) through CEI at 12 mL/h or through demand-dose PCEA. Patients using PCEA could demand 3 mL every 10 min without restriction. Analgesia in both groups was comparable. However, there was a significant reduction in total bupivacaine consumption (in milligrams) associated with the use of PCEA (mean +/- SEM: CEI = 76.1 +/- 8.5 mg; PCEA = 42.2 +/- 5.9 mg; 45% reduction). The hourly bupivacaine consumption during the first (CEI = 15.8 +/- 0.6 mg/h; PCEA = 8.8 +/- 1.1 mg/h) and second (CEI = 17.2 +/- 1.2 mg/h; PCEA = 6.8 +/- 1.2 mg/h) stages of labor was also reduced. Overall, this represented a 47% "sparing" of bupivacaine use per hour with PCEA. Similar reductions occurred in the use of fentanyl. The reductions in analgesic requirement, however, were not associated with a reduction in the degree of motor blockade or in the cephalad extent of sensory blockade. A significant dose-sparing effect was associated with the use of demand-dose PCEA as compared with standard CEI for analgesia during labor and delivery.