With a view to strengthening systems for maternal health, this study presents a comprehensive analysis of determinants of uptake of care and pregnancy-related mortality, with the main emphasis on distance to care. Data on geographical positioning (GIS), socio-demographic information, birth histories and deaths in women of reproductive age were collected during a household census in five rural districts of Southern Tanzania in 2007. Deaths reported as pregnancy-related were followed up by verbal autopsies. Health facility census information collected in the same area in 2009 was used. Data limitations included 30% either missing or low quality GIS data and missing birth histories for 9% of women. The analysis included 507 pregnancy-related deaths and 64,098 live births. Major deficiencies in quality of care provided in health facilities were identified. Although 75% of women lived within a distance of 4.6km to a facility providing delivery care, overall institutional delivery was low with 29% of all births in hospital and 11% in first-line facilities. Seventy-two percent of women living <5km away delivered in hospital and levels declined rapidly thereafter with no evidence of confounding. In contrast, less than 30% of women delivered in a first-line facility even if they lived less than 1km away. Overall pregnancy-related mortality was high at 712 deaths per 100,000 livebirths (95% Confidence Interval 652-777), with 32% due to haemorrhage. There was weak evidence of higher mortality with increasing distance to hospital, which was accentuated if the analysis was restricted to direct maternal deaths. Sensitivity analysis restricting analysis to the 70% of households with good quality GIS data did not alter conclusions. There was no evidence that low uptake of care at first-line facilities was explained by distance or socio-demographic factors. Deficiencies in quality of care influence both care uptake and mortality suggesting that investments in quality should be prioritized.