Tamoxifen, a rat liver carcinogen, can induce mutations in the lacI gene in the livers of lambda/lacI transgenic rats. However, the presence of persistent tamoxifen adducts on the liver DNA raises the possibility that some contribution to the mutagenesis from ex vivo mutations during the in vitro lacI assay cannot be ruled out. To address this issue, mutagenesis at the cII gene of the transgenic shuttle vector was determined using a selection based assay which is unaffected by the presence of tamoxifen-DNA adducts. Female lambda/lacI transgenic rats were dosed orally with tamoxifen (20 mg/kg body wt) daily for 6 weeks, causing a 3.2-fold increase in the mutant frequency (MF) in the cII gene compared with that obtained with solvent treated animals. This was similar to the MF found previously at the lacI gene and confirms that tamoxifen is mutagenic in vivo. The major class of mutation induced by tamoxifen in the cII gene was G:C-->T:A transversions as was found previously in the lacI gene. However, in the one unreplicated study of mutations in the p53 gene of liver tumours induced by tamoxifen, no G:C-->T:A transversions were found; possible differences between mutagenesis in normal and tumour tissues are explored. The major proportion of the G:C-->T:A transversions occurred at 5'-CpG-3' dinucleotide (CpG) sites in the lacI gene, but not at such sites in the cII gene. The methylation of CpG sites greatly enhances the targeting of deoxyguanosine by carcinogens, thus this finding might be explained by differences in the methylation patterns at their respective CpG sites; however, nothing is known about the methylation status of either the lacI nor the cII gene in this transgenic rat. This study raises the important issue of which target genes (mammalian or transgenic) should be used as endpoints in mammalian mutagenesis assays.