A seroepidemiologic, nested case-control study was conducted to evaluate the risk for atrophic gastritis associated with Helicobacter pylori infection. Atrophic gastritis was diagnosed on the basis of serum pepsinogen levels: pepsinogen I level < or = 70 ng/ml and pepsinogen I/pepsinogen II ratio < or = 3.0. Cases were 23 men and 39 women who were not diagnosed with atrophic gastritis in 1987, but who were diagnosed with the condition in 1992. Controls were the 120 men and 282 women who did not meet the serologic criteria for atrophic gastritis in either time period. Neither cases nor controls had a history of upper gastrointestinal operations. Helicobacter pylori infection at the initial survey was associated with a significantly increased risk of atrophic gastritis incidence for both sexes combined (odds ratio = 3.72; 95% confidence interval, 1.78-7.79; P = 0.0005). Cigarette smoking and consumption of alcohol and green-yellow vegetables were not associated with incidence of atrophic gastritis.