To investigate whether using low dose epidural infusion improves the normal delivery rate, outcome of labour was studied in women with singleton vertex presentations randomised to receive either 0.0625% bupivacaine opioid, or plain bupivacaine 0.125% for labour. The infusion rate was titrated to maintain analgesia and a sensory level to T10. Data were analysed using the unpaired t test, Mann-Whitney U test and for categorical variables chi2 test. Adjusted odds ratios for factors significantly associated with non-normal delivery were calculated using stepwise logistic regression. There were 291 women in the low dose and 296 in the plain bupivacaine group. There were no significant differences between groups in parity, race, induction of labour, use of augmentation, cervical dilatation at epidural insertion, duration of any stage of labour or duration or volume of infusion. Total dose of bupivacaine (126 +/- 47 mg versus 91 +/- 32 mg) and the proportion of women with motor block at the end of labour (45% versus 27%) were significantly greater in the plain bupivacaine than in the low dose group (P < 0.0001). The adjusted odds ratios (95% CI) for factors significantly associated with non-normal delivery were primiparity: 4.68 (2.78-7.88), older maternal age: 1.1 (1.05-1.14), longer active second stage of labour: 1.01 (1.005-1.017), total bupivacaine dose: 1.01 (1.005-1.016) and greater cervical dilatation at epidural insertion 1.22 (1.08-1.37). Treatment group and motor block at the end of labour had no significant effect. We found no increase in normal delivery rate with low dose infusions.