A study was undertaken to determine the role of Helicobacter pylori in non-ulcer dyspepsia (NUD) and to determine the efficacy of colloidal bismuth subcitrate (CBS) in the treatment of NUD. Seventy-one patients were randomly allocated (double blind) to CBS or placebo, two tablets twice daily for 4 weeks. The severity of dyspepsia was scored and endoscopies performed before and after treatment, and antral biopsy specimens were taken for bacteriologic and histologic examination. Forty patients had H. pylori infection, and all had changes of chronic active gastritis. H. pylori was cleared from 17 to 21 patients (81%) treated with CBS, whereas none of the 19 patients treated with placebo cleared the bacteria. Improvement in histology was noted in 15 of 21 patients (71.4%) treated with CBS, whereas no improvement was noted in any of the placebo controls. Thirty-one patients were negative for H. pylori. All had either normal gastric histology or minor degrees of inflammation. Seventeen of these patients received CBS, and 14 received placebo. All groups reported improvement in the symptom score; however, the H. pylori-positive, CBS-treated group recorded a significantly higher improvement than the other groups (p less than 0.001). Relapse of H. pylori infection after initial clearance of the bacteria was high. Twelve of 16 patients evaluated relapsed 1 month after withdrawal of CBS.