By means of hospital-based data over 9 years we sought to evaluate the clinical indications and incidence of emergency peripartum hysterectomy by demographic characteristics and reproductive history. From the obstetric records of all deliveries at Brigham and Women's Hospital between Oct. 1, 1983, and July 31, 1991, we identified all women undergoing emergency peripartum hysterectomy, calculated crude and adjusted incidence rates, conducted statistical tests of linear trends and heterogeneity, and observed the clinical indications preceding the onset of this procedure. There were 117 cases of peripartum gravid hysterectomy identified during this period, for an overall annual incidence of 1.55 per 1000 deliveries. The rate increased with increasing parity and was significantly influenced by placenta previa and a history of cesarean section. The incidence by parity increased from one in 143 deliveries in nulliparous women with placenta previa to one in four deliveries in multiparous women with four or more deliveries with placenta previa. Likewise, the incidence increased from one in 143 deliveries in women with one prior live birth and a prior cesarean section to one in 14 deliveries in multiparous women with four or more deliveries with a history of a prior cesarean section. Both these trends were highly significant (p < 0.001). Abnormal adherent placentation was the most common cause preceding gravid hysterectomy (64%, p < 0.001), with uterine atony accounting for 21%. Although no maternal deaths occurred, maternal morbidity remained high, including postoperative infection in 58 (50%), intraoperative urologic injury in 10 patients (9%), and need for transfusion in 102 patients (87%). The data identify abnormal adherent placentation as the primary cause for gravid hysterectomy. The data also illustrate how the incidence of emergency peripartum hysterectomy increases significantly with increasing parity, especially when influenced by a current placenta previa or a prior cesarean section. Maternal morbidity remained high although no maternal deaths occurred.