We investigated the role of dosing regimen on ras mutations in chemically induced CD-1 mouse liver tumors. The spectra of ras gene mutations in liver tumors that were induced by 15 daily i.p. injections of 7,12-dimethylbenz[a]anthracene (DMBA), 4-aminoazobenzene (AAB), N-hydroxy-2-acetylaminofluorene (N-OH-AAF) or N-nitrosodiethylamine (DEN) were compared to those previously obtained for tumors induced by a single but higher dose of each carcinogen. The principal assay used was a direct tumor analysis involving sequencing of polymerase chain reaction (PCR)-amplified tumor DNA; additional mutations that were present in only a small fraction of tumor cells were detected using a transfection assay or a PCR-engineered restriction fragment length polymorphism method. Spontaneous liver tumors had a relatively low frequency of ras mutations, all found in Ha-ras codon 61, and most of these mutations were present in only a small fraction of tumor cells. With the exception of multiple-dose DEN, each group of single- and multiple-dose carcinogen-induced tumors exhibited a higher frequency of ras mutations compared with spontaneous tumors. For AAB, N-OH-AAF and DEN, the dosing regimen was found to affect significantly the profile of ras mutations. For each of these carcinogens, the multiple-dose tumor group (versus single-dose group) had fewer Ki-ras and N-ras mutations and more tumors in which the Ha-ras codon 61 (C-->A) mutation was present in a large fraction of cells. Our results demonstrate that the dosing procedure can materially affect the pattern of ras gene mutation in mouse liver tumors.