Although retained placenta is a major cause of postpartum hemorrhage, there is no general agreement regarding when manual placental extraction is indicated to prevent hemorrhage. We sought to determine the following: 1) what duration of the third stage of labor is abnormal, 2) what duration is associated with complications, and 3) what antecedent conditions are associated with prolonged third stage. We studied 12,979 consecutive, singleton vaginal deliveries over an 11-year period. Third-stage duration had a log-normal distribution, with a geometric mean of 6.8 minutes, a median of 6 minutes, and an interquartile range of 4-10 minutes. A third stage of 30 minutes or longer occurred in 3.3% of the deliveries. The incidence of postpartum hemorrhage, transfusion, and D&C remained constant in third stages less than 30 minutes, then rose progressively, reaching a plateau at 75 minutes. The increase in these complications after 30 minutes was observed with both spontaneously delivered and manually extracted placentas. In a logistic regression analysis, factors significantly associated with prolonged third stage included: preterm delivery (odds ratio 3.81), delivery in a labor bed (odds ratio 2.17), preeclampsia (odds ratio 1.76), augmented labor (odds ratio 1.47), and nulliparity (odds ratio 1.45). Because there was no increase in hemorrhage until the third stage exceeded 30 minutes, we suggest that in the absence of bleeding, manual placental extraction is not indicated until 30 minutes have elapsed.