BACKGROUND: Gemeprost and misoprostol are two of the most widely used prostaglandins in combination with mifepristone for medical abortion in early pregnancy. However, the efficacy and side-effects of those two drugs given vaginally have not been assessed in a randomized trial. METHODS: Randomized double-blind controlled trial involving 999 women undergoing an abortion at gestational age < or =63 days who received either 0.5 mg gemeprost (group I, n = 499) or 800 microg misoprostol (group II, n = 500) vaginally approximately 48 h after taking 200 mg mifepristone by mouth. The rate of complete abortion and the side-effects were compared between the groups. RESULTS: A total of 89 cases was excluded from full analysis of outcome because either they aborted after mifepristone alone (n = 2), had an ectopic pregnancy (n = 1), or because the outcome was uncertain as they failed to attend their follow-up appointment (n = 86). The rate of complete abortion was very high (>95%) in both groups but significantly higher after treatment with misoprostol than with gemeprost [436/453 (98.7%) versus 451/457 (96.2%), P = 0.019, difference 2.5%, confidence interval 0.4-4.7%] and there were fewer ongoing pregnancies (n = 1 versus n = 8, P < 0.018). Surgical intervention rose significantly with gestation in women who received gemeprost (P < 0.03) but not with misoprostol. The incidence of side-effects such as diarrhoea (13.7 versus 16.4%) and vomiting (27.8 versus 29.7%) was similar in women who received misoprostol or gemeprost respectively, as was the duration and amount of bleeding. CONCLUSIONS: (i) Both regimens using a reduced dose of mifepristone are highly effective methods of inducing abortion in early pregnancy; (ii) vaginal misoprostol is the preferred prostaglandin because it is it is associated with fewer failures than low-dose gemeprost, particularly at gestation > or =49 days.