OBJECTIVE: to explore the relationship between the degree to which labour is established on admission to hospital and method of birth. BACKGROUND: a recent randomised controlled trial found fewer caesarean sections (CS) in women allocated to caseload midwifery (19.4%) compared with standard care (24.9%). There is interest in exploring what specific aspects of the care might have resulted in this reduction. SETTING: a large tertiary-level maternity service in Melbourne, Australia. PARTICIPANTS: English-speaking women with no previous caesarean section at low risk of complications in pregnancy were recruited to a randomised controlled trial. Trial participants whose management did not include a planned caesarean and who were admitted to hospital in spontaneous labour were included in this secondary analysis of trial data (n=1532). METHODS: this secondary analysis included women admitted to hospital in spontaneous labour who were randomised to caseload midwifery compared with those randomised to standard care with regard to timing of admission in labour, augmentation of labour and use of epidural analgesia. In a further analysis randomised groups were pooled to examine predictors of caesarean section for first births only using multiple logistic regression. RESULTS: nulliparous women randomised to standard care were more likely to have labour augmented than those having caseload care (54.2% and 45.5% respectively, p=0.008), but were no more likely to use epidural analgesia. They were admitted earlier in labour, spending 1.1 hours longer than those in the caseload arm in hospital before the birth (p=0.003). Parous women allocated to standard care were more likely than those in the caseload arm to use epidural analgesia (10.0% and 5.3% respectively, p=0.047), but were no more likely to have labour augmented. They were also admitted earlier in labour, with a median cervical dilatation of 4 cm compared with 5 cm in the caseload arm (p=0.012). Pooling the two randomised groups of nulliparous women, and after adjusting for randomised group, maternal age and maternal body mass index, early admission to hospital was strongly associated with caesarean section. Admission before the cervix was 5 cm dilated increased the odds 2.4-fold (95%CI 1.4, 4.0; p=0.001). Augmentation of labour and use of epidural analgesia were each strongly associated with caesarean section (adjusted odds ratios 3.10 (95%CI 2.1, 4.5) and 5.77 (95%CI 4.0, 8.4) respectively. CONCLUSION: these findings that women allocated to caseload care were admitted to hospital later in labour, and that earlier admission was strongly associated with birth by caesarean section, suggest that remaining at home somewhat longer in labour may be one of the mechanisms by which caseload care was effective in reducing caesarean section in the COSMOS trial.