Chronic infection with the human bacterial pathogen Helicobacter pylori causes gastritis and predisposes carriers to an increased gastric cancer risk. Consequently, H. pylori-specific vaccination is widely viewed as a promising strategy of gastric cancer prevention. H. pylori strains harboring the Cag pathogenicity island (PAI) are associated with particularly unfavorable disease outcomes in humans and experimental rodent models. We show in this study using a C57BL/6 mouse model of Cag-PAI(+) H. pylori infection that the only known protein substrate of the Cag-PAI-encoded type IV secretion system, the cytotoxin-associated gene A (CagA) protein, harbors MHC class II-restricted T cell epitopes. Several distinct nonoverlapping epitopes in CagA's central and C-terminal regions were predicted in silico and could be confirmed experimentally. CagA(+) infection elicits CD4(+) T cell responses in mice, which are strongly enhanced by prior mucosal or parenteral vaccination with recombinant CagA. The adoptive transfer of CagA-specific T cells to T cell-deficient, H. pylori-infected recipients is sufficient to induce the full range of preneoplastic immunopathology. Similarly, immunization with a cholera toxin-adjuvanted, CagA(+) whole-cell sonicate vaccine sensitizes mice to, rather than protects them from, H. pylori-associated gastric cancer precursor lesions. In contrast, H. pylori-specific tolerization by neonatal administration of H. pylori sonicate in conjunction with a CD40L-neutralizing Ab prevents H. pylori-specific, pathogenic T cell responses and gastric immunopathology. We conclude that active tolerization may be superior to vaccination strategies in gastric cancer prevention.