Objective: to compare intrapartum- and neonatal mortality and intervention rates in term women starting labour in primary midwife-led versus secondary obstetrician-led care. Design: retrospective cohort study. Setting: Amsterdam region of the Netherlands. Participants: women with singleton pregnancies who gave birth beyond 37+0 weeks gestation in the years 2005 up to 2008 and lived in the catchment area of the neonatal intensive care units of both academic hospitals in Amsterdam. Women with a primary caesarean section or a pregnancy complicated by antepartum death or major congenital anomalies were excluded. For women in the midwife-led care group, a home or hospital birth could be planned. Measurements: analysis of linked data from the national perinatal register, and hospital- and midwifery record data. We assessed (unadjusted) relative risks with confidence intervals. Main outcome measures were incidences of intrapartum and neonatal (<28 days) mortality. Secondary outcomes included incidences of caesarean section and vaginal instrumental delivery. Findings: 53,123 women started labour in primary care and 30,166 women in secondary care. Intrapartum and neonatal mortality rates were 37/53,123 (0.70‰) in the primary care group and 24/30,166 (0.80‰) in the secondary care group (relative risk 0.88; 95% CI 0.52-1.46). Women in the primary care group were less likely to deliver by secondary caesarean section (5% versus 16%; RR 0.31; 95% CI 0.30-0.32) or by instrumental delivery (10% versus 13%; RR 0.76; 95% CI 0.73-0.79). Key conclusions: we found a low absolute risk of intrapartum and neonatal mortality, with a comparable risk for women who started labour in primary versus secondary care. The intervention rate was significantly lower in women who started labour in primary care. Implications for practice: these findings suggest that it is possible to identify a group of women at low risk of complications that can start labour in primary care and have low rates of medical interventions whereas perinatal mortality ...