Understanding the strengths and limitations of alternative models, such as the Tg.AC assay, for evaluation of the potential carcinogenicity of pharmaceuticals requires assessment of assay specificity through studies that specifically target biologically active compounds that are known to not be carcinogens in rodents. To identify drugs that might provoke a false positive response in the Tg.AC assay, we screened pharmaceuticals for in vitro induction of the gadd153 promoter and the zeta-globin promoter. We have previously found a high correlation between induction of the gadd153 promoter in HepG2 cells and activity in the Tg.AC assay. The three drugs selected through screening 99 noncarcinogenic pharmaceuticals were amiloride, dipyridamole, and pyrimethamine. A 26-week skin paint study was conducted in hemizygous Tg.AC mice with the three drugs at two doses selected by a 4-week dose range finding study. Evidence of systemic toxicity was observed in animals dosed chronically with pyrimethamine or amiloride, but no skin papillomas were observed in mice treated with amiloride, dipyridamole, or pyrimethamine for 26 weeks. All male mice and 80% of female mice treated with 12-O-tetradecanoylphorbol-13-acetate (TPA) in acetone developed a maximal tumor burden. However, mice treated with TPA in a vehicle containing 2.4% DMSO had greatly reduced incidences of papillomas. In summary, the correct negative response was shown in the Tg.AC assay for three noncarcinogenic pharmaceuticals, which adds further favorable evidence of appropriate specificity of this model system. However, vehicle composition must be carefully selected because the outcome of this assay can be confounded by certain commonly used solvents.