Our purpose was to compare the safety and effectiveness of prostaglandin E2 delivered sequentially as an intracervical (0.5 mg) or intravaginal (2.5 mg) gel. Hospitalized patients with an unfavorable cervix (Bishop score < or = 4) at > or = 35 weeks and requiring induction of labor were assigned to receive two 2.5 ml doses of gel intracervically and intravaginally in a double-blind, placebo-controlled manner. Second and third doses were given at 6-hour intervals until there were either regular uterine contractions or a Bishop score change > 3 points. The 100 evaluable cases received prostaglandin E2 either intracervically (n = 52) or intravaginally (n = 48). Difficulty with exact gel instillation was present with intracervical gel only, where spillage occurred in 85% of cases. Compared with intracervical therapy prostaglandin E2 given intravaginally was more likely to significantly change the Bishop score (60.4% vs 40.4%, p = 0.04) and stimulate regular contractions (72.9% vs 48.1%, p = 0.01). Uterine hyperstimulation was present in one case in each group. Although each was safe, instillation of prostaglandin E2 gel was better at a higher intravaginal dose than a lower intracervical dose because of its greater ease of administration and higher likelihood of cervical change.