The definition of postpartum hemorrhage (PPH) was developed more than 50 years ago. Since then, the obstetric population has changed dramatically. We sought to determine how well we estimated blood loss (EBL) and find thresholds predicting need for transfusion.We performed a prospective cohort study from 2010 to 2012, comparing those who needed transfusion postpartum and those who did not. EBL, calculated EBL (cEBL), and change in hematocrit were calculated for patients who did not receive transfusion, and EBL was calculated for those who did receive transfusion, stratified by delivery type. Receiver operator curves were created and optimal thresholds of EBL to predict transfusion were estimated.Among 4,804 patients, transfusion was required for 0.65% of vaginal and 8.7% of cesarean deliveries. Median EBL was higher in women requiring transfusion. A weak correlation was noted between EBL and cEBL for all deliveries. Thresholds of 500 mL blood loss for vaginal delivery and 1,000 mL for cesarean had the best predictive ability for transfusion.In this modern obstetric, cohort EBL is weakly correlated with cEBL, suggesting that accuracy of clinical estimates of blood loss is modest. However, EBL predicts need for transfusion, with optimal thresholds of 500 mL for a vaginal delivery and 1,000 mL in a cesarean. This validates the traditional definitions of PPH in our modern population.