This study determined the optimal maintenance dose of omeprazole in reflux oesophagitis. One hundred and ninety three patients rendered asymptomatic and healed after four or eight weeks omeprazole were randomised double blind to 10 mg omeprazole once daily (n = 60 evaluable), 20 mg omeprazole once daily (n = 68), or placebo (n = 62) for one year or until symptomatic relapse. Each omeprazole regimen was superior to placebo in preventing both symptomatic relapse (life table analysis, p < 0.001) and endoscopically verified relapse (p < 0.001). At 12 months, the life table endoscopic remission rates (proportions of patients without grade > or = 2 oesophagitis) were: 50% (95% confidence intervals 34 to 66%) with 10 mg omeprazole once daily, 74% (62 to 86%) with 20 mg omeprazole once daily, and 14% (2 to 26%) with placebo. At 12 months, the life table symptomatic remission rates (proportions of patients asymptomatic or with mild symptoms) were: 77% (64 to 89%) with 10 mg omeprazole once daily, 83% (73 to 93%) with 20 mg omeprazole once daily, and 34% (16 to 52%) with placebo. Both 10 mg and 20 mg omeprazole once daily were effective in prolonging the remission of reflux oesophagitis: 10 mg may be appropriate to start longterm treatment, though the existence of a dose response relation means that 20 mg once daily may be effective in patients for whom 10 mg once daily is suboptimal.