Ethylene oxide (EO) is a genotoxicant and a mouse lung carcinogen, but whether EO is carcinogenic through a mutagenic mode of action remains unclear. To investigate this question, 8-week-old male Big Blue B6C3F₁ mice (10 mice/group) were exposed to EO by inhalation-6 h/day, 5 days/week for 4 weeks (0, 10, 50, 100, or 200 ppm EO) and 8 or 12 weeks (0, 100, or 200 ppm EO). Lung DNA samples were analyzed for levels of 3 K-ras codon 12 mutations (GGT→GAT, GGT→GTT, and GGT→TGT) using ACB-PCR. No measureable level of K-ras codon 12 TGT mutation was detected (ie, all lung mutant fractions [MFs] ≤ 10⁻⁵). Four weeks of inhalation of 100 ppm EO caused a significant increase in K-ras codon 12 GGT→GTT MF relative to controls, whereas 50, 100, and 200 ppm EO caused significant increases in K-ras codon 12 GGT→GAT MF. In addition, significant inverse correlations were observed between K-ras codon 12 GGT→GTT MF and cII mutant frequency in the lungs of the same mice exposed to 50, 100, or 200 ppm EO for 4 weeks. Surprisingly, 8 weeks of exposure to 100 and 200 ppm EO caused significant decreases in K-ras MFs relative to controls. Thus, the changes in K-ras MF as a function of cumulative EO dose were nonmonotonic and were consistent with EO causing early amplification of preexisting K-ras mutations, rather than induction of K-ras mutation through genotoxicity at codon 12. The possibility that these changes reflect K-ras mutant cell selection under varying degrees of oxidative stress is discussed.