OBJECTIVE The study's objectives were as follows: (1) to determine the rate of vaginal delivery after labor induction in severe preeclampsia remote from term and (2) to determine potential predictors of success. METHODS Retrospective chart review was conducted on live-born singleton pregnancies complicated by severe preeclampsia and delivered at 24 to 34 weeks' gestation from January 1, 1992, to December 31, 1996. Exclusion criteria included eclampsia, presence of labor or spontaneous rupture of membranes on admission, and complication of pregnancy by an ultrasonographically detected fetal congenital anomaly. Patients were divided into 3 groups: elective cesarean delivery without labor, cesarean delivery after labor induction, and vaginal delivery after labor induction. Statistical analyses included multiple logistic regression, the Student t test, the chi2 test, and the Mann-Whitney test. P </=.05 was considered significant. RESULTS A total of 306 charts were reviewed. Among these, 161 patients (52.6%) underwent elective cesarean delivery without labor; the 2 most common indications were unfavorable cervix (33.5%) and malpresentation (22.4%). The remaining 145 patients (47.4%) underwent labor induction with a 48. 3% rate of vaginal delivery after induction, ranging from 31.6% at </=28 weeks' gestation to 62.5% at >32 weeks' gestation. The most common indication for cesarean delivery after induction, in 50.7% of the cases, was nonreassuring fetal heart rate. The median Bishop score was significantly higher (3 vs 2, P =.004) and the total hospital stay was significantly shorter in the vaginal delivery after induction group than in the cesarean delivery after induction group. However, there were no significant differences between the 2 groups in use of cervical ripening agents, gestational age at delivery, birth weight, 5-minute Apgar score, or postpartum endometritis. After exclusion of cesarean deliveries performed for malpresentation, there was no statistically significant difference in classic incision rates between the elective cesarean delivery without labor and cesarean delivery after induction groups (13.6% vs 6.8%; P =.137). According to logistic regression analysis, only the Bishop score was significantly associated with a successful induction (odds ratio 1.38, 95% confidence interval 1.11-1.71). Gestational age reached marginal significance (odds ratio 1.30, 95% confidence interval 0.89-1.89). CONCLUSIONS (1) Labor induction should be considered a reasonable option for patients with severe preeclampsia at </=34 weeks' gestation because 48% of patients given the chance were successfully delivered vaginally. (2) The Bishop score on admission is the best predictor of success, although the chance of successful labor induction increases with advancing gestational age.