Background and Aim: Experimental carcinogenesis models provide a useful tool in the study of the aetiopathogenesis and treatment of gastric cancer. We developed a model based on the administration of N-methyl-N-nitrosourea (NMU) in Wistar rats for the induction of maximal yield of gastric carcinomas with a short latency period, and being exclusively localized at the gastric level. Methods: A gastric antiperistaltic fistula was performed in 90 Wistar rats classified into eight different groups. Fifteen days after surgery 5, 10, 15 or 20 mg of NMU/100 g were administered through the fistula once a week for a 3- to 5-week period. Before the administration of NMU, a pyloric blockade was made in order to obtain a temporary isolation of the stomach. At 20 weeks, animals were sacrificed and organs were removed for histological study. Results: All rats treated with 15 mg NMU/100 g once a week for 5 weeks, after pyloric blockade maintained for 1 h, developed well-differentiated carcinomas in the forestomach. Carcinomas were multiple in 11% of cases and appeared with papillomatous lesions in 33% of rats. No tumours were observed in any other organs. In the other groups, no gastric carcinomas were diagnosed. Conclusion: The high incidence of carcinomas in the forestomach, the absence of tumours in other organs and the short latency period represent valuable criteria for the use of our model in chemotherapeutic investigations, as well as in the study of cancer evolution without interferences caused by tumour development in other organs.