The aim of this study was to determine the effect of colloidal bismuth subcitrate (De Nol) on symptoms and gastric histology in patients with non-ulcer dyspepsia. In a single centre trial, patients with food related upper abdominal pain not caused by ulcer disease were randomised to receive one tablet of colloidal bismuth subcitrate or matching placebo four times daily for eight weeks. Seventy three patients were entered and 51 completed the trial: 28 patients in the colloidal bismuth subcitrate group and 23 in the placebo group. Overall there was no difference between the two groups in terms of symptom relief. Among patients with histological gastritis (n = 23), however, those who took colloidal bismuth subcitrate used fewer antacid tablets (for three of four fortnightly periods) and were more likely to become asymptomatic (eight of 11 v three of 12, p less than 0.05); their gastritis was more likely to resolve (five of 10 v 0 of 12, p less than 0.025) and their gastric biopsies more likely to become negative for Helicobacter like organisms (eight of nine v 0 of 12, p less than 0.001) when compared with patients taking placebo. In contrast, patients who did not have gastritis in their index biopsies (n = 28) fared similarly whether they received colloidal bismuth subcitrate or placebo. Our results indicate that the administration of colloidal bismuth subcitrate benefited non-ulcer dyspepsia patients with gastritis but had no effect on those without.