Many mutations in the p53 gene destroy the transcriptional transactivation function of the p53 protein. This function of p53 can be determined in a yeast assay using a p53 responsive reporter gene. The yeast assay could hold promise for the identification of mutagens implicated in human cancer if the p53 mutational spectra obtained with this assay would match human tumor mutation data. Ultraviolet (UV) light from the sun and polycyclic aromatic hydrocarbons, such as benzo[a]pyrene, are strongly implicated in the spectrum of p53 mutations found in human non-melanoma skin cancers and smoking-associated lung cancers, respectively. We have used these two model mutagens to assess the feasibility of using the p53 yeast assay in cancer epidemiology. After treatment of CpG methylated p53 DNA with a solar UV simulator or with benzo[a]pyrene diol epoxide (BPDE), the modified p53 sequences were assayed in yeast for mutational outcome. As expected, BPDE produced predominantly G to T transversions and simulated sunlight produced mostly C to T transitions at dipyrimidine sites in the p53 coding sequence. However, the preferentially mutated p53 sequences (hotspots) in the yeast assay were completely different from those in the mutational spectra found in human lung and skin cancers. The data indicate that this assay is not a reliable measurement of p53 mutagenesis in human tissues and that, perhaps, transcriptional activation is not the primary function of p53 in tumor suppression.