We have mapped sites of tamoxifen adduct formation, in the lacI gene using the polymerase STOP assay, following reaction in vitro with alpha-acetoxytamoxifen and horseradish peroxidase (HRP)/H(2)O(2) activated 4-hydroxytamoxifen. For both compounds, most adduct formation occurred on guanines. However, one adenine, within a run of guanines, generated a strong polymerase STOP site with activated 4-hydroxytamoxifen, and a weaker STOP site with alpha-acetoxytamoxifen at the same location. In Escherichia coli the lac I gene reacted with 4-hydroxytamoxifen was more likely to be mutated (2 orders of magnitude) than when reacted with alpha-acetoxytamoxifen, despite the greater DNA adduct formation by alpha-acetoxytamoxifen. This correlates with the greater predicted ability of activated 4-hydroxytamoxifen adducts to disrupt DNA structure than alpha-acetoxytamoxifen adducts. For lac I reacted with activated 4-hydroxytamoxifen, a hot spot of base mutation was located in the region of the only adenosine adduct. No mutational hot spots were observed with alpha-acetoxytamoxifen. Our data clearly shows a lack of correlation between gross adduct number, as assayed by (32)P-postlabeling and mutagenic potential. These data indicate the importance of minor adduct formation in mutagenic potential and further that conclusions regarding the mutagenicity of a chemical may not be reliably derived from the gross determination of adduct formation.