At present, there is no study that simultaneously addresses the apparent differences between bacterial and host factors in patients with bleeding and nonbleeding Helicobacter pylori-related ulcer diseases. Therefore, we designed this prospective study to evaluate whether there are identifiable differences between the two groups of patients whose H. pylori-related peptic ulcer diseases present with bleeding or dyspepsia. From July 1996 to November 1996, consecutive patients presenting with upper gastrointestinal bleeding or dyspepsia were enrolled if H. pylori-related ulcer diseases were confirmed. Fifteen clinical, endoscopic, histologic, and serologic factors were tested for association with ulcer bleeding by a logistic regression analysis. In the study period, bleeding occurred in 39 out of 119 patients with H. pylori-related peptic ulcer diseases. Multivariate analysis showed that ingestion of nonsteroidal antiinflammatory drugs (NSAIDs; p = 0.0156; odds ratio = 5:4), ulcer size > or = 1 cm (p = 0.0033; odds ratio = 4:2), and low bacterial density (p = 0.0030; odds ratio = 4:1) were independent factors associated with the risk of bleeding. There were no associations between ulcer bleeding and age, sex, smoking, alcohol consumption, the histologic grade of gastritis, location and number of ulcers, and the cytotoxin-associated gene (CagA) status of H. pylori strain. Therefore, we concluded that H. pylori-related ulcer patients who use NSAIDs or have large ulcers are more likely to present with upper gastrointestinal bleeding; that the CagA-bearing strains are not associated with the development of bleeding complication in patients with peptic ulcer diseases; and that the exact reason concerning the association between low bacterial density and ulcer bleeding merits further investigation.