The Testin (TES) gene was previously identified as a putative human tumor suppressor gene at 7q31.2, a region that is frequently deleted in hematopoietic malignancies, as well as in epithelial tumors. To determine whether TES acts as a tumor suppressor in vivo, we generated a Tes knockout mouse and then used it in an established model of carcinogen-induced gastric cancer. In mice a zinc-deficient (ZD) diet enhances cellular proliferation in the forestomach and susceptibility to N-nitrosomethylbenzylamine (NMBA)-induced carcinogenesis. Five-week-old Tes wild-type (+/+), heterozygous (+/-), and homozygous (-/-) mice were divided into four groups: mice fed a zinc-sufficient diet (ZS); mice fed a ZD diet; ZS fed plus NMBA-treated mice (ZS+NMBA), and ZD fed plus NMBA-treated mice (ZD+NMBA). After 4 weeks, the ZS+NMBA and ZD+NMBA groups were treated with three intragastric doses of NMBA. Animals were killed 8 weeks after NMBA administration: 25% of +/+ mice developed benign lesions; 88% of +/- showed multiple papillomas, atypical glandular metaplasia, and squamous cell carcinomasl; and 81% of -/- mice displayed very large papillomas, squamous cell carcinomas, and adenocarcinomas. A statistically significant difference in tumor incidence was found between +/- versus +/+ and -/- versus +/+ (P < 0.0001). These data suggest that Tes functions as a tumor suppressor gene in vivo.