A prospectively designed review of all obstetric hysterectomies performed in five university hospitals between November 1, 1984 and October 31, 1987 has been performed. There were 41,107 deliveries and 46 obstetric hysterectomies, an incidence of 0.11%. Twenty-five hysterectomies were elective and 21 were emergent. The indication for 11 of the 21 emergency hysterectomies was placenta previa and/or accreta. Women in the emergency group had greater intraoperative blood loss, were more likely to have intraoperative hypotension, and were more likely to receive donor blood than women in the elective group (P less than 0.05). Twelve patients (eight from the elective group and four from the emergency group) received continuous epidural anesthesia, and none required intraoperative induction of general anesthesia. There was no evidence that epidural anesthesia significantly affected blood loss, crystalloid replacement, or requirement for transfusion in the elective group. Abnormal placentation now represents a major indication for emergency obstetric hysterectomy. Furthermore, significant hemorrhage is more likely with emergency obstetric hysterectomy than with elective hysterectomy. Finally, elective cesarean hysterectomy is not a contraindication to performance of continuous epidural anesthesia.