OBJECTIVE: Reducing the rate of severe postpartum haemorrhage (PPH) is a major challenge in obstetrics today. One potentially effective tool for improving the quality of care is the clinical audit, that is, peer evaluation and comparison of actual practices against explicit criteria. Our objective was to assess the impact of regular criteria-based audits on the prevalence of severe PPH. DESIGN: Quasi-experimental before-and-after survey. SETTING: Two French maternity units in the Rhône-Alpes region, with different organization of care. PARTICIPANTS: All staff of both units. INTERVENTION: Quarterly clinical audit meetings at which a team of reviewers analysed all cases of severe PPH and provided feedback on quality of care and where all staff actively participated. MAIN OUTCOME MEASURES: The primary outcome was the prevalence of severe PPH. Secondary outcomes included the global quality of care for women with severe PPH, including the performance rate for each recommended procedure. Differences in these variables between 2005 and 2008 were tested. RESULTS: The prevalence of severe PPH declined significantly in both units, from 1.52 to 0.96% of deliveries in the level III hospital (P = 0.048) and from 2.08 to 0.57% in the level II hospital (P < 0.001). From 2005 to 2008, the proportion of deliveries with severe PPH that was managed consistently with the guidelines increased for all of its main components, in both units. CONCLUSION: Regular clinical audits of cases severe PPH were associated with a persistent reduction in the prevalence of severe PPH.