Objective: To analyze the accuracy of postpartum hemorrhage risk factors to determine patients at risk of severe postpartum hemorrhage and transfusion. Population and methods: Retrospective cohort study from a database in one high-risk obstetric unit over a 7-year period. Results: In a cohort of 19,204 deliveries, 44 patients were transfused of whom five were given frozen fresh plasma only. Of the 39 who received red blood cells, 35 received at least three units. Multivariate analysis of postpartum hemorrhage risk factors revealed a significant role of placenta previa/accreta, cesarean section, multiple pregnancy, prematurity and vascular disease. Nevertheless 28% of women transfused had none of these risk factors. Conclusion: The percentage of patients transfused has probably decreased markedly with improved prevention, surveillance and treatment. This study emphasizes that the transfusion risk in the presence of anomalous placental insertion justifies special obstetrical and anesthetic management.