We report on the determinants and consequences of induced abortion among 803 women admitted to hospital with abortion complications in Rio de Janeiro, Brazil, in 1991. 458 (57%) women reported using misoprostol to induce abortion, 74% in the first 4 months of pregnancy. Doses of 200-16,800 micrograms were reported, with a median of 800 micrograms. 65% of the women took the drug orally, 29% used a combination of oral and vaginal routes, and 6% administered it intravaginally. Vaginal bleeding and uterine cramps were the commonest reasons for seeking hospital care. Only 8% of women reported gastrointestinal side-effects. Misoprostol induced vaginal bleeding within 12 h of administration in 52% of the women, but 16% waited 10 days or more for onset of bleeding. 4% were admitted to hospital with complete abortion. The likelihood of bleeding starting within 12 h increased with duration of gestation and it was greater when the drug was used both orally and intravaginally. A significantly smaller proportion of women taking misoprostol than of those who induced abortion by catheter insertion presented signs of infection or physical injuries or required blood transfusion (< 0.0005). Among 803 women interviewed at delivery as controls, 6% had taken misoprostol but abortion had not ensued. Misoprostol has an important role as an abortifacient among the women studied.