One hundred patients with duodenal ulceration and Campylobacter pylori in their stomach were entered into a double blind placebo controlled prospective study. Treatment schedules were cimetidine and placebo, or cimetidine and tinidazole, or colloidal bismuth subcitrate (CBS) and placebo, or CBS and tinidazole. Seventeen per cent of isolates of C pylori obtained at the first endoscopy were resistant to tinidazole and 70% of the second isolates from patients given cimetidine and tinidazole became tinidazole resistant. Suspensions of nitroimidazole sensitive cultures of C pylori showed that three of 22 isolates had a nitroimidazole resistant subpopulation. In patients who healed and remained free of C pylori after treatment ulcers recurred less often than in patients who healed but retained C pylori (23% v 73% over 12 months, p < 0.001).