A longitudinal cohort study was conducted in Helicobactor pylori-infected middle-aged Japanese males to evaluate the preventive effects of H. pylori eradication on the development of gastric cancer according to the extent of chronic atrophic gastritis (CAG). The extent of CAG was monitored by baseline serum pepsinogen (PG) levels. We followed 3,656 subjects with persistent H. pylori infection and 473 subjects with successful H. pylori eradication for cancer development for a mean (SD) of 9.3 (0.7) years. Groups with and without extensive CAG were categorized based on PG test-positive criteria to detect extensive CAG of PG I <or= 70 ng/ml and PG I/II ratio <or= 3.0. During the study period, 5 and 55 gastric cancers developed in H. pylori-eradicated and the noneradicated subjects, respectively, indicating no significant reduction in cancer incidence after H. pylori eradication. Among the noneradicated subjects, 1,329 were PG test-positive and 2,327 were PG test-negative. Gastric cancer was confirmed in 30 and 25 subjects, respectively. Among subjects whose infection was eradicated, 155 were PG test-positive and 318 were PG test-negative. Of these subjects, gastric cancer was confirmed in 3 and 2 subjects, respectively. Significant reduction in cancer incidence after eradication was observed only in PG test-negative subjects (p < 0.05; log-rank test). The results of this study strongly indicate that cancer development after eradication depends on the presence of extensive CAG before eradication and that H. pylori eradication is beneficial to most PG test-negative subjects with mild CAG as defined by the aforementioned criteria.