We investigated a possible outbreak of H. pylori in a rural Northern Plains community. In a cross-sectional survey, we randomly sampled 244 households from a geocoded emergency medical system database. We used a complex survey design and global positioning system units to locate houses and randomly selected one eligible household member to administer a questionnaire and a 13C-urea breath test for active H. pylori infection (n = 166). In weighted analyses, active H. pylori infection was detected in 55·0% of the sample. Factors associated with infection on multivariate analysis included using a public drinking-water supply [odds ratio (OR) 12·2, 95% confidence interval (CI) 2·9–50·7] and current cigarette smoking (OR 4·1, 95% CI 1·7–9·6). People who lived in houses with more rooms, a possible indicator of decreased crowding in the home, were less likely to have active H. pylori infections (OR 0·7, 95% CI 0·5–0·9 for each additional room).