Helicobacter pylori infection has been associated with gastritis, duodenal ulcer, gastric ulcer, and the epidemic form of gastric carcinoma. Eradication of H. pylori infection has proven to be difficult. Recently, combinations of antimicrobial drugs have been shown to eradicate greater than 50% of infections; however, the results have proven variable, and the factors influencing effectiveness of therapy are unclear. In the present study, the effectiveness of a triple therapy for eradication of H. pylori infection was evaluated. Triple therapy consisted of 2 g tetracycline, 750 mg metronidazole, and five or eight tablets of bismuth subsalicylate daily in 93 patients (70 with duodenal ulcer, 17 with gastric ulcer, and 6 with simple H. pylori gastritis). Combinations of a sensitive urea breath test, serology, culture, and histology were used to confirm the presence of infection, eradication, or relapse. Eradication was defined as inability to show H. pylori greater than or equal to 1 month after ending therapy. The overall eradication rate was 87%. The factors evaluated for their effect on predicting eradication included age, gender, type of disease, duration of therapy, amount of bismuth subsalicylate [five or eight Pepto-Bismol tablets daily (Procter & Gamble, Cincinnati, OH)], and compliance with the prescribed medications. Stepwise regression showed that compliance was the most important factor predicting success; the success rate was 96% for patients who took greater than 60% of the prescribed medications and 69% for patients who took less. For those taking greater than 60% of the prescribed therapy, the eradication rates were similar (a) for patients receiving therapy for 14 days or when tetracycline and bismuth subsalicylate were taken for an additional 14 days; (b) for patients with duodenal ulcer, gastric ulcer, and simple H. pylori gastritis; and (c) whether five or eight bismuth subsalicylate tablets were taken. It is concluded that triple therapy is effective for eradication of H. pylori and that future studies need to take compliance into account for comparisons between regimens.