A multicenter, double-blind, randomized study compared 200 micrograms of misoprostol and placebo four times daily for four weeks in the treatment of 225 patients with duodenal ulcer (0.7 cm to 2.0 cm in size) persisting after at least four weeks of adequate, conventional therapy with cimetidine or ranitidine. Misoprostol was significantly superior to placebo in healing duodenal ulcers (achieving a healing rate of 37 percent versus 22 percent in the placebo group [p = 0.02], and in relieving ulcer pain [p = 0.01]). Healing also occurred more frequently with misoprostol than with placebo in patients with subgroups of particularly resistant ulcers. In the treatment of ulcers refractory to at least eight weeks of histamine H2-blocker therapy, misoprostol achieved a healing rate of 42 percent versus 20 percent with placebo. In the treatment of pyloric channel ulcers, 28 percent of patients in the misoprostol group showed healing as compared with 20 percent in the placebo group. Diarrhea was reported by 15.4 percent and 3.4 percent of patients receiving misoprostol and placebo, respectively, and was usually mild and transient. Misoprostol is safe and effective therapy for duodenal ulcers that have not healed during the course of H2-blocker therapy.