To identify symptoms and risk factors associated with Helicobacter pylori infection, a cohort of 341 epidemiologists was studied. All subjects had one banked serum (collected between 1969 and 1987) and one recent serum sample (collected in 1988) evaluated for H. pylori immunoglobulin G by enzyme-linked immunosorbent assay; subjects provided information on gastrointestinal symptoms and risk factors for gastritis and peptic ulcer disease. Prevalence of infection decreased from the early 1970s to the present. Eleven subjects (3% of the total cohort) seroconverted during the interval between serum samples, giving a crude conversion rate of 0.49% per person-year (95% confidence interval, 0.3-0.9). Nonreactors on the 1988 serum sample described similar symptoms to reactors. However, subjects who seroconverted in the interval between the two serum samples were more likely than either persistent nonreactors [relative risk (RR), 4.1] or persistent reactors (RR, 3.7) to have experienced upper gastrointestinal symptoms in the interval years. Consumption of caffeinated beverages (RR, 4.6) and residence in the northeastern United States (RR, 5.3) seemed to increase risk for infection. Because pain was similarly common in H. pylori-positive and -negative patients, H. pylori cannot be summarily accepted as the cause of dyspeptic symptoms even when infection is confirmed.