A case-control study was performed to study risk factors for postpartum hemorrhage. Cases of hemorrhage were defined by a hematocrit decrease of 10 points or more between admission and post-delivery or by the need for red-cell transfusion. Patients with antenatal bleeding were excluded. Among 9598 vaginal deliveries, postpartum hemorrhage occurred in 374 cases (3.9%). Three controls were matched to each case and multiple logistic regression was used to control for covariance among predictor variables. Factors having a significant association with hemorrhage were prolonged third stage of labor (adjusted odds ratio 7.56), preeclampsia (odds ratio 5.02), mediolateral episiotomy (4.67), previous postpartum hemorrhage (3.55), twins (3.31), arrest of descent (2.91), soft-tissue lacerations (2.05), augmented labor (1.66), forceps or vacuum delivery (1.66), Asian (1.73) or Hispanic (1.66) ethnicity, midline episiotomy (1.58), and nulliparity (1.45). These data may help predict postpartum hemorrhage and may be useful in counseling patients about the advisability of home delivery, intravenous access in labor, or autologous blood donation.