Background The induction of full-term labour in women with a live fetus remains a major challenge in modern obstetrics.Objectives To determine, using the best level of evidence available, the efficacy and safety of sublingual administration of misoprostol compared with vaginal misoprostol in the third trimester of pregnancy for the induction of labour, according to initial doses, in women with a live, full-term fetus and an unripe cervix.Search strategy Pubmed/Medline, Lilacs and Scielo databases were consulted, as well as clinical trials registered in the Cochrane Register from January 1996 to February 2008, using the keywords 'misoprostol', 'labour, obstetric', 'delivery, obstetric', 'induced labour' and 'parturition' with the search limits of 'clinical trials' and 'randomised clinical trials'.Selection criteria This review contains randomised clinical trials in which the sublingual and vaginal routes of administration of misoprostol were compared. Participants were pregnant women with an indication for induction of labour and a live fetus more than 37 weeks of gestational age.Data collection and analysis The primary analysis compared sublingual and vaginal routes of administration of misoprostol. Secondary analyses compared different routes and initial doses of misoprostol. Statistical analysis included odds ratios and their respective 95% CI. To evaluate the heterogeneity of the studies, the I-squared test was used, studies being considered heterogeneous when I 2 was greater than 50%.Main results Five good quality clinical trials involving a total of 740 women were eligible, and all were included. No statistically significant difference was found between the sublingual and the vaginal misoprostol groups with respect to the rate of vaginal delivery not achieved within 24 hours (OR 1.27, 95% CI 0.87-1.84), uterine hyperstimulation syndrome (OR 1.20, 95% CI 0.61-2.33) or caesarean section (OR 1.33, 95% CI 0.96-1.85). An increased risk of uterine tachysystole was found in the sublingual misoprostol group (OR 1.70, 95% CI 1.02-2.83). When the studies were grouped according to the initial dose of misoprostol, no significant difference was found between sublingual or vaginal groups.Author's conclusions The sublingual route of administration is as effective as the vaginal route in inducing labour in full-term pregnancies with live fetuses. However, the safety, adverse effects, optimal dose and perinatal outcome related to this route of administration remain to be established, and it cannot be recommended for routine use in obstetric practice.