BACKGROUND: Cohort and case-crossover studies were conducted to evaluate whether new Helicobacter pylori infections are followed by increased diarrhea. METHODS: Participants were 6-month-old to 12-year-old shantytown residents living near Lima, Peru. Baseline data were collected from community households. Health interviews were completed daily, and sera, drawn every 4 months, were tested for H pylori immunoglobulin G. Diarrhea rates among newly H pylori-infected (seroconverting) children were compared with rates among persistently uninfected and infected children using cohort and case-crossover analyses. RESULTS: Sera were obtained from 345 children from January 1, 1995, through September 1, 1997. H pylori incidence was 12% per year (36 H pylori infections in 109 866 seronegative days). In adjusted cohort analyses, seroconverters had more diarrhea days (rate ratio: 2.0; 95% confidence interval: 1.6-2.4), episodes, and sick days in the year after infection than did uninfected children; and more diarrhea days and sick days than did persistently infected children. This effect was strongest in the first 2 months. Case-crossover analyses supported these findings. CONCLUSION: Preventing H pylori infection may help reduce pediatric diarrheal disease.