Oxazepam has been the subject of recent toxicological and carcinogenesis studies because it is a commonly prescribed tranquilizer and has been shown to cause tumors in rodents. In this study, male and female B6C3F1 mice received 0, 125, 2500 or 5000 p.p.m. oxazepam in the diet for up to 2 years. Hepatocellular adenomas and carcinomas, as well as hepatoblastomas, which developed in these mice, were examined for the presence of activated ras proto-oncogenes. DNA was isolated from 20 or more tumors from each exposure group and analyzed by oligonucleotide hybridization, single-stranded conformation polymorphism analysis and direct sequencing of PCR-amplified H-ras gene fragments for codon 61 mutations. Thirteen of 37 (35%) hepatocellular adenomas and carcinomas from the 125 p.p.m. exposure group had mutations in codon 61, while mutations were detected in only 2 of 25 or 8% of the liver tumors from the 2500 p.p.m. exposure group and none of the 22 tumors from the 5000 p.p.m. group. This compares to 63% of 126 historical control liver tumors and 55% of 20 liver tumors from unexposed B6C3F1 mice in this study. In addition, 12 hepatoblastomas from the two high dose groups were examined for H-ras mutations at codon 61, but none were detected. No tumor DNAs from any of the exposure groups tested had mutations in codons 12, 13 or 117 of the H-ras gene or codons 12 or 13 of the K-ras gene, the other known hotspots for ras activation in mouse liver tumors. These results, together with those from the National Toxicology Program study showing no evidence of cytotoxicity or genotoxicity by oxazepam, suggest that oxazepam preferentially promotes cells that have activating lesions other than ras.