We have studied the response of erosive or ulcerative esophagitis to treatment with omeprazole and its subsequent relapse on cessation of therapy in 196 patients. In the first phase of the study omeprazole (20 or 40 mg daily) was compared with placebo in 64 patients. After 4 wk there was endoscopic healing in 81% (25 of 31) of omeprazole-treated patients and in only 6% (2 of 32) of placebo-treated patients. Endoscopic healing of esophagitis was accompanied by symptom relief and histologic healing of ulceration. In the second (dose finding) phase a further 132 patients were randomized to omeprazole (20 or 40 mg daily) and endoscopic healing was assessed. In patients with the mildest grade of ulcerative esophagitis (grade 2), healing occurred at 4 wk in 87% receiving 20 mg and in 97% receiving 40 mg. In patients with grade 3 esophagitis, 67% (20 mg) and 88% (40 mg) were healed. Less than half the patients with grade 4 esophagitis (Barrett's ulcers or confluent ulceration) healed with either 20 mg (48%) or 40 mg (44%). Regression analysis in the 164 omeprazole-treated patients showed no evidence that healing was influenced by factors other than severity of esophagitis at entry and omeprazole dose. In phase 3 of the study the rate of endoscopic relapse was determined in 107 endoscopically healed patients after stopping omeprazole. Erosive or ulcerative esophagitis recurred in 88 of 107 (82%) by 6 mo. Neither initial dose, grade of esophagitis, nor smoking was shown to influence relapse rate. Omeprazole is a highly effective treatment for peptic esophagitis. The 40-mg/day dosage produces endoscopic healing slightly more quickly than the 20-mg/day dosage, and the initial endoscopic gradings are of prognostic value. Relapse occurs rapidly when treatment is stopped.