BACKGROUND: Few studies have assessed whether the poorest people in developing countries benefit from giving birth at home rather than in a facility. We analysed whether socioeconomic status results in differences in the use of professional midwives at home and in a basic obstetric facility in a rural area of Bangladesh, where obstetric care was free of charge. METHODS: We routinely obtained data from Matlab, Bangladesh between 1987 and 2001. We compared the benefits of home-based and facility-based obstetric care using a multinomial logistic and binomial log link regression, controlling for multiple confounders. FINDINGS: Whether or not a midwife was used at home or in a facility differed significantly with wealth (adjusted odds ratio comparing the wealthiest and poorest quintiles 1.94 [95% CI 1.69-2.24] for home-based care, and 2.05 [1.72-2.43] for facility-based care). The gap between rich and poor widened after the introduction of facility-based care in 1996. The risk ratio (RR) between the wealthiest and poorest quintiles was 1.91 (adjusted RR 1.49 [95% CI 1.16-1.91] when most births with a midwife took place at home compared with 2.71 (1.66 [1.41-1.96]) at the peak of facility-based care. INTERPRETATION: In this area of Bangladesh, a shift from home-based to facility-based basic obstetric care is feasible but might lead to increased inequities in access to health care. However, there is also evidence of substantial inequities in home births. Before developing countries reinforce home-based births with a skilled attendant, research is needed to compare the feasibility, cost, effectiveness, acceptability, and implications for health-care equity in both approaches.