BACKGROUND: Access to skilled obstetric delivery and emergency care is deemed crucial for reducing maternal mortality. We assessed pregnancy-related mortality by distance to health facilities and by cause of death in a disadvantaged rural area of southern Tanzania. METHODS: We did a secondary analysis of cross-sectional georeferenced census data collected from June to October, 2007, in five rural districts of southern Tanzania. Heads of georeferenced households were asked about household deaths in the period June 1, 2004, to May 31, 2007, and women aged 13-49 years were interviewed about birth history in the same time period. Causes of death in women of reproductive age were ascertained by verbal autopsy. We also asked for sociodemographic information. Multilevel logistic regression was used to analyse the effects of distance to health facilities providing delivery care on pregnancy-related mortality (direct and indirect maternal and coincidental deaths). FINDINGS: The study included 818 583 people living in 225 980 households. Pregnancy-related mortality was high at 712 deaths per 100 000 livebirths, with haemorrhage being the leading cause of death. Deaths due to direct causes of maternal mortality were strongly related to distance, with mortality increasing from 111 per 100 000 livebirths among women who lived within 5 km to 422 deaths per 100 000 livebirths among those who lived more than 35 km from a hospital (adjusted odds ratio 3·68; 95% CI 1·37-9·88). Neither pregnancy-related nor indirect maternal mortality was associated with distance to hospital. Among women who lived within 5 km of a hospital, pregnancy-related mortality was 664 deaths per 100 000 livebirths even though 72% gave birth in hospital and 8% had delivery by caesarean section. INTERPRETATION: Large distances to hospital contribute to high levels of direct obstetric mortality. High pregnancy-related mortality in those living near to a hospital suggests deficiencies in quality of care. FUNDING: Bill & Melinda Gates Foundation.