Objective：To elucidate the major reasons for unexpected death nude mice when they were used for establishing human metastatic cancer models. Methods： The fresh specimens of human renal cell carcinoma （RCC）, prostate cancer （PC）, and DU-145 cells were transplanted / injected into nude mice ecotopically and orthotopically. Tumorigenesis and pathological changes （including the symptoms, pathological sections, survival time, etc. ） of mouse liver were investigated subsequently. Results： The tumorigenesis and metastasis rates were respectively 21.7% （35/161） and 1.2% （2/161） after implantation of RCC sample, and were respectively 100V0 （20/20） and 25M （5/20） after implantation of DU-145 cell line, while there was no tumorigenesis or metastasis after implantation of PC specimens. Liver pathological changes were found in 58. 4%（94/161） of mice implanted with RCC samples and in 43.4% （46/ 106） of mice implanted with PC samples. No pathological lesion was found in mice implanted with DU-145 cells. The death peak of mice with pathological changes after implanting RCC and PC samples was consistent with that of the total mice used in this study, all oceurring in the winter and spring of the year. Conclusion： The pathological changes of liver appear to be the major reason of unexpected death of the nude mice when they were used for establishing human metastatic cancer models. A specified pathogen-free environment is very important for establishment of the models