Objective To determine which interventions for managing placenta accreta were associated with reduced maternal morbidity.Design Retrospective cohort study.Setting Two tertiary care teaching hospitals in Utah.Population All identified cases of placenta accreta from 1996 to 2008.Methods Cases of placenta accreta were identified using standard ICD-9 codes for placenta accreta, placenta praevia, and caesarean hysterectomy. Medical records were then abstracted for maternal medical history, hospital course, and maternal and neonatal outcomes. Maternal and neonatal complications were compared according to antenatal suspicion of accreta, indications for delivery, preoperative preparation, attempts at placental removal before hysterectomy, and hypogastric artery ligation.Main outcome measures Early morbidity (prolonged maternal intensive care unit admission, large volume of blood transfusion, coagulopathy, ureteral injury, or early re-operation) and late morbidity (intra-abdominal infection, hospital re-admission, or need for delayed re-operation).Results Seventy-six cases of placenta accreta were identified. When accreta was suspected, scheduled caesarean hysterectomy without attempting placental removal was associated with a significantly reduced rate of early morbidity compared with cases in which placental removal was attempted (67 versus 36%, P = 0.038). Women with preoperative bilateral ureteric stents had a lower incidence of early morbidity compared with women without stents (18 versus 55%, P = 0.018). Hypogastric artery ligation did not reduce maternal morbidity.Conclusions Scheduled caesarean hysterectomy with preoperative ureteric stent placement and avoiding attempted placental removal are associated with reduced maternal morbidity in women with suspected placenta accreta.