We studied 645 full-term low-risk women in early labour in 6 units to evaluate the effects of maternal characteristics and obstetric management in early labour on the use of epidural analgesia, and to analyse the relationship between epidural analgesia, progress of labour and mode of delivery using multiple logistic regression. Among variables present in early labour, nulliparity, ethnicity and obstetric unit were the strongest predictors of epidural analgesia requirement. In nulliparous women, obstetric unit affected use of epidural analgesia (P<0.05) and induction of labour was associated with increased use of epidural analgesia (odds ratio 3.45, 95% CI: 1.45-7.90). In multiparous women, only ethnicity was statistically significant (P<0.05). Epidural analgesia was associated with longer labours and more instrumental deliveries (odds ratio 2.93, 95%CI: 1.48-5.83). In the epidural group, however, we found a positive correlation between first stage duration and elapsed time before epidural analgesia. Furthermore, rate of cervical dilation was similar in the non epidural group throughout the first stage (mean 3.41 cm/h, 95%CI: 3.19-3.63) and in the epidural group after epidural analgesia decision (mean 3.99, 95% CI: 2.96-5.02), while the mean cervical dilatation rate before epidural analgesia was 0.88 cm/h (95% CI: 0.72-1.04). The need for epidural analgesia is, therefore, multifactorial and difficult to predict. Whereas nulliparity increases epidural analgesia requirement, data on the progress of labour before pain relief suggest that epidural analgesia is a marker of pain severity and/or labour failure rather than the cause of delayed progress in low-risk pregnancies.