Objective To evaluate the stability of a compound modeling method of chronic atrophic gastritis with N-methyl-N’-nitro-N-nitrosoguanidine（ MNNG） and its effect on animal hepatocytes. Methods Sixty specificpathogen-free healthy male Sprague Dawley rats were randomly divided into a normal group（ n = 8）

a model group 1（ n = 26） and a model 2 group（ n = 26）. The model group 1 adopted MNNG combined with high salt hot starch paste method and the model 2 group adopted MNNG combined with ethanol gavage method to prepare rat model with chronic atrophic gastritis. The normal group received normal feeding. Rats underwent modeling for twenty eight weeks. Gastric mucosa pathological degree

serum levels of pepsinogen Ⅰ（ PG Ⅰ） and pepsinogen Ⅱ（ PG Ⅱ）

and the change of liver pathology of rats in the model group 1 and the model 2 group were compared after modeling. Results Rats in the model group 1 had gland hyperplasia at the 24 week

and had focal intestinal metaplasia at the 26 week.Rats in the model 2 group had individual gland hyperplasia at the 26 week

and had focal intestinal metaplasia at the28 week.

The mortality rate was 26. 9% in the model 1 group

and 15. 4% in the model 2 group. After modeling

there was no statistically significant difference in gastric mucosa pathological changes between rats in the model 1 and in the model 2 group（ P > 0. 05）. Compared with those in the normal group

serum levels of PGⅠ

and PGⅠ/PGⅡ in the model 1 group and model group 2 decreased（ P < 0. 05）. At the 28 week of modeling

hepatocytes of both rats in the model 1 group and model group 2 appeared hydropic degeneration. Conclusion The MNNG compound modeling method of establishing model of chronic atrophic gastritis needs longer modeling time. It is with high mortality

and has a certain effect on the liver tissue of rats.