BACKGROUND: This study explored the reasons for variation in hospital maternal mortality ratio (MMR) between studies from sub-Saharan Africa. METHODS: A systematic review was conducted to identify hospital-based studies which reported the prevalence of maternal mortality. An overall MMR from all the hospital-based studies was calculated using a metaanalysis. Potential sources of heterogeneity in the MMR between studies were identified using metaregression techniques. RESULTS: We identified 4243 studies, of which 64 were eligible for inclusion in the metaanalysis. The pooled hospital MMR for sub-Saharan Africa was 957 per 100 000 live births, although there was strong evidence for between-study heterogeneity. Regional estimates varied from 294 per 100 000 live births in Southern Africa to 1338 in Western Africa. Overall, throughout the region, the percentage of skilled birth attendance and type of hospital accounted for 44% of the total variation of the hospital MMR between studies. CONCLUSION: This paper highlights the need to improve the organisation of health systems and the quality of care that is being offered in health facilities to pregnant women in Africa; and emphasizes the importance of increasing the percentage of skilled birth attendance in the region. In order to achieve the Millennium development goal (MDG) and reduce maternal mortality in the region, particularly in Western Africa, new and stronger approaches are needed.