OBJECTIVE: To examine the nature of the association between maternal mortality and birth with a health professional in observational studies. METHODS: Review of ecological studies relating the overall proportion of births with a health professional with the maternal mortality ratio at national level, and studies exploring the relationship between the presence of a health professional at birth and the risk of dying at the individual level. We report methodological challenges, including data quality and sources and the analytical approaches used. For the individual studies, crude odds ratios and 95% confidence intervals were calculated. RESULTS: The 10 ecological studies are largely descriptive, a causal inference is tentative and there is poor controlling of confounders. The 10 individual studies examining the risk of death with and without a health professional showed little evidence that giving birth with a health professional reduces a woman's risk of dying, and in some settings it appears to be associated with an increased risk of death. CONCLUSIONS: None of these study designs are optimal in evaluating the impact of births with a health professional on reducing maternal mortality. Analytically, greater insights can be gained by examining ecological relationships within countries, and by complementing the individual analyses with information on the health status of women when they first reach the health professional and whether or not the women planned to have a health professional present during birth.