Objective: To explore the feasibility of 7, 12-dimethylbenz[a] anthracene (DMBA) induced tree shrew breast cancer model, and compare the effects of two administration modes by gavage and mammary gland injection. Methods: A total of 40 tree shrews were randomly divided into two groups (20 animals per group): DMBA gavage group and mammary gland injection group. DMBA was dissolved in edible vegetable oil. For gavage group, tree shrews were administered with DMBA solutions (15 mg/kg) by gavage once a day. For mammary gland injection group, DMBA solution (10 mg/kg) was injected into the mammary fat pad of tree shrews, and the injection was performed for a total of 3 times. From the first administration of DMBA, medroxyprogesterone acetate (MPA, 100 mg/kg) was intramuscularly injected into the muscles of the lateral thighs of tree shrews at the same time, for a total of 5 times. The tumorigenesis and survival of tree shrews were monitored. The tumor histological morphology was observed by HE staining. The expression of estrogen receptor (ER), progesterone receptor (PR), cytokeratin5/6 (CK5/6) and human epidermal factor receptor-2 (HER-2) was detected by immunohistochemical staining. Results: In the gavage group, there were 10 deaths, and 4 tree shrews developed mammary tumors with 20.0% (4/20) tumor formation rate. The success rate of mammary cancer modeling was 10.0% (2/20), and the tumor formation time was 197.3±15.1 days. In the mammary gland injection group, there were 8 tree shrews died, and 9 tree shrews formed tumors with 45.0% (9/20) tumor formation rate. The success rate of mammary cancer modeling was 40.0% (8/20), and the tumor formation time was 71.8±19.0 days. There was no significant difference in mortality and tumor formation rate (P>0.05) between the two groups (all P>0.05). However, in the mammary gland injection group, the success rate of mammary cancer modeling was significantly higher than that in the gavage group (P<0.05), whereas the tumor formation time was markedly shorter than that in the gavage group (P<0.01). The pathological types in the gavage group included ductal hyperplasia, intraductal papilloma and ductal carcinoma in situ, while those in the breast injection group included intraductal papilloma and ductal carcinoma in situ. In both groups, immunohistochemical staining showed the negative expression of HER-2 but positive expression of ER, PR and CK5/6 with varying degrees. Conclusion Both the DMBA gavage and mammary gland injection can successfully establish the tree shrew breast cancer model, and the modeling effect of mammary gland injection is better than gavage.