Introduction: Although obstetric hemorrhage is one of the most common causes of maternal death worldwide, its prevalence is poorly understood. The objective of this population-based study was to assess the utility of PMSI data for estimating the prevalence of obstetric hemorrhage in France. Method: We linked the PMSI with the database of the French Blood Establishment (EFS, our reference database) for the Burgundy region, then examined the discordant cases. Based on this study, we developed a multivariate regression model to correct the total number of pregnant women identified in the national PMSI as having received a transfusion. This study was conducted using data from the Burgundy region and included all parturients for the 2006–2007 period (35,799 pregnancies). The modeling covered all births in France in 2006 and 2007 (1,629,537 pregnancies). Results: the sensitivity of the PMSI in detecting pregnant women who received a transfusion was estimated at 66.3% with a positive predictive value of 91.3%. The total number of pregnant women identified in the PMSI as having received a transfusion at the national level was estimated, after correction, at 10,941 for 2006–2007, or 6.71%. Conclusion: The proposed method allows for a corrected estimate of the frequency of obstetric hemorrhage at the national level. (R.A.)