Comparison of the mutation patterns of p53 in human tumors with those of selectable genes in model systems is a powerful approach to identify potential etiological factors for specific tumor types. Recently, we validated use of a yeast assay to permit direct determination of the mutation spectrum induced in human p53 by carcinogens that would reduce uncertainties inherent in comparing spectra induced in different target genes. Here, we describe modifications in the assay designed to facilitate screening for mutants and to permit intracellular exposure of the gene instead of in vitro treatment. This was accomplished by introducing growth-based selection for transactivation-deficient p53 mutants into yeast already possessing red/white colony color selection. This improved model system was able to detect cells harboring p53 mutations among cells with wild-type p53 at a frequency of 10(-4) or less. Additionally, UV light was used to verify that the majority of mutagenized cells with the appropriate phenotype on selective medium contained mutations in p53, not elsewhere in the genome. Sequence analysis of UV-induced mutations revealed that the nature of the mutations was similar to those obtained in previous studies of this mutagen. This system will prove useful in the determination of the ability of environmental agents to mutate the human p53 gene, and thus may contribute to hazard identification.