A randomised, single-blind study was conducted among 52 gravida in active labour, to investigate two variants of patient-controlled epidural analgesia--bolus only versus bolus plus infusion. Patient-controlled analgesia variables, using an epidural solution of 0.125% bupivacaine plus fentanyl 3 mcg per ml, were a 4 ml incremental bolus with 15 minute lockout, plus or minus a 4 ml per hour infusion. Up to three additional staff-administered supplements of 0.5% bupivacaine 4 ml (20 mg) were allowed. There was no significant difference between groups with respect to pain relief, supplementary boluses required, satisfaction, side-effects or details of patient-controlled epidural analgesia, with the exception of greater fentanyl usage in the bolus plus infusion group (P less than 0.003). Both groups had high quality analgesia, low rates of bupivacaine usage and were highly satisfied. However, under the conditions of the study, the addition of a continuous background infusion to self-administration conferred no benefit.