Mouse models are important tools in toxicologic research. Differences between species in pathways contributing to tumor development, however, raise the question in how far mouse models are valid for human risk assessment. One striking difference relates to the frequency of spontaneous liver cancer which is high in certain mouse strains but rather low in humans. Similarly, mutation frequencies in cancer genes are characteristically different, i.e. P53 mutations are frequent in human but very rare in murine liver tumors, whereas Ras genes are often mutated in mouse liver tumors but hardly ever in human liver cancers. Since P53 has been shown to control oncogenic RAS in human cells, we hypothesized that this function of the tumor suppressor could differ in mouse hepatocytes. To test this hypothesis, we used hupki (human p53 knock-in) mice which carry a partly humanized P53 sequence (P53KI). In this study, we report the results of the first hepatocarcinogenesis experiment with this strain of mice. Mice of the genotypes P53KI/KI, P53WT/KI and P53WT/WT were treated with N-nitrosodiethylamine at 2 weeks of age and killed 35 weeks later. The frequency of liver tumors and glucose-6-phosphatase-altered liver lesions was almost identical in all three P53 genotypes and approximately 40-50% of liver tumors showed activating mutations in codon 61 of the Ha-Ras gene independent of genotype. Moreover, only very few P53-positive lesions were observed but without nuclear localization of the protein, suggesting the absence of P53 mutations. These data suggest that the hupki allele behaves like its murine ortholog in mouse hepatocarcinogenesis.