In this double-blind, parallel-group multicenter study, patients with endoscopically proven gastric ulcers were randomly allocated to treatment with either 50 g or 200 g of misoprostol or 300 mg of cimetidine, each given four times daily for four weeks. Endoscopic, clinical and laboratory assessments were made before treatment and after four weeks; clinical and laboratory assessments were repeated at two weeks. In the Korean center, assessments were also made after six weeks and at eight weeks of treatment. Six hundred and thirty patients were studied. The three treatment groups were similar in age and occupation. However, the proportion of men in the misoprostol 50-g, 200-g and cimetidine 300-mg groups was 67%, 63%, and 59%, respectively. Therapeutic success was defined as complete healing of all ulcers, judged endoscopically. On an intent-to-treat basis, which includes all losses to follow-up and withdrawals as treatment failures, ulcer healing rates in the misoprostol 50-g, 200-g and cimetidine 300-mg groups were 39%, 51%, and 58%, respectively. In the Korean center, the healing rates were 38%, 64%, and 70%, respectively, after eight weeks of treatment. There was no statistically significant difference in the healing rates at four weeks between the misoprostol 200-g and cimetidine 300-mg groups (P=0.16). The healing rate with the misoprostol 200-g dose was significantly better than with the 50-g dose (P=0.008). Cimetidine 300 mg relieved global pain significantly better than misoprostol 200 g at two weeks (P=0.047) but not at four weeks. The 200-g dose of misoprostol relieved pain significantly better than the 50-g dose at four weeks (P=0.019), but not at two weeks. All three treatments were well tolerated. Severe adverse events were rare.