Ninety-five healthy nulliparous women, ASA physical status I-II with an uncomplicated pregnancy and single fetus in vertex position were given lumbar epidural analgesia. Patients in Group A (n = 35) received bupivacaine 0.125 per cent with epinephrine 1:800.000; Groups B (n = 30) and C (n = 30) received the same agents as Group A but with the addition to the initial dose of 50 or 100 micrograms of fentanyl respectively. All patients were evaluated for duration and quality of analgesia, duration of labour, method of delivery and total dose of bupivacaine used. The addition of either 50 or 100 micrograms of fentanyl resulted in longer duration of analgesia (93 +/- 9 min and 106 +/- 8 min respectively vs 55 +/- 7) and reduced bupivacaine total doses (64 +/- 0.03 and 55 +/- 1.5 respectively vs 109.5 +/- 1.3). Only the addition of 100 micrograms of fentanyl improved significantly the quality of analgesia (43.3 per cent of excellent scores vs 6.6 per cent in Group B and 5.7 per cent in Group A). Addition of fentanyl did not affect the duration of labour, the method of delivery and the neonatal neurobehaviour scores.