Objective: To determine which of three methods of cervical ripening resulted in the lowest cesarean rate in women with unfavorable cervices and indications for labor induction. Methods: Consenting women with singleton gestations, vertex presentations, and unfavorable cervices (dilatation under 2 cm and effacement under 75%) were randomly assigned to laminaria and standard intravenous oxytocin, serial doses of intracervical prostoglandin (PG) E2 gel (Prepidil, Pharmacia and Upjohn, Inc., Kalamazoo, MI) 0.5 μg every 6 hours for two doses followed by oxytocin if indicated, or extra-amniotic saline infusion and oxytocin. Results: An interim analysis after recruitment of 321 subjects, 67% of the planned sample, found similar cesarean rates for the three groups (laminaria 36%; PGE2 gel 33%; saline infusion 29%; P = .59); however, the mean randomization-to-delivery interval was significantly longer in the PGE2 group. Stochastic curtailment, as part of the interim analysis, indicated a low likelihood of achieving a statistically significant difference in cesarean rates between PGE2 gel and the other two groups. Therefore, we completed the study with saline infusion and laminaria. The saline infusion and laminaria groups had similar preinduction characteristics. The cesarean rates were similar (saline infusion 25.4% versus laminaria 30.3%; P = .32), but the mean interval from randomization to delivery was shorter in the saline infusion group (18.0 versus 21.5 hours, P = .002). There were no significant differences in selected maternal and neonatal morbidities. Conclusion: Cervical ripening with extra-amniotic saline infusion, PGE2, or laminaria resulted in comparable cesarean rates in women with an unfavorable cervix and indications for labor induction. Extra-amniotic saline infusion had the shortest randomization-to-delivery interval without increasing maternal or neonatal morbidity. Copyright (C) 2000 The American College of Obstetricians and Gynecologists.