Liver tumors from mice treated with genotoxic carcinogens often possess mutations in ras protooncogenes, and these sequence alterations in ras frequently reflect the mutational specificity of the carcinogen. Previous studies suggest that the mouse model used for tumor induction may affect ras mutational patterns. In order to explore this possibility, H- and K-ras mutational profiles were established in liver tumors from male B6C3f1 and CD-1 mice administered benzo[a]pyrene (BaP), 6-nitrochrysene (6-NC), and 4-aminobiphenyl (4-ABP). With the exception of 6-NC-induced tumors in B6C3F1 mice, a high proportion of the tumors induced in both types of mice contained ras mutations. In CD-1 mice, 6-NC predominantly induced C-->A mutations in H-ras codon 61 (90% of tumors analyzed), whereas 4-ABP mainly induced A-->T mutations in H-ras codon 61 (50%) and BaP induced both A-->T (27%) and G--> (50%) mutations in H-ras codon 61 and K-ras codon 13, respectively. In B6C3F1 mice, 85% of BaP tumors had G-->C mutations in K-ras codon 13 and 85% of 4-ABP tumors had C-->A mutations in H-ras codon 61, while among 6-NC tumors, only 4% had G-->C mutations in K-ras codon 13 and none had H-ras mutations. Statistical analysis of these results indicates that the patterns of tumor ras mutations induced by BaP in CD-1 and B6C3F1 mice were indistinguishable, while 6-NC and 4-ABP produced different tumor ras profiles in the two mouse models. Published mutational profiles for active metabolites of BaP and 6-NC from in vitro reporter gene systems were inconsistent with both the CD-1 and B6C3F1 tumor ras mutational responses.