It has been shown that Smad3 exerts both tumor-suppressive and -promoting roles. To evaluate the role of Smad3 in skin carcinogenesis in vivo, we applied a chemical skin carcinogenesis protocol to Smad3 knockout mice (Smad3^−/− and Smad3^+/−) and wild-type littermates (Smad3^+/+). Smad3^−/− mice exhibited reduced papilloma formation in comparison with Smad3^+/+ mice and did not develop any squamous cell carcinomas. Further analysis revealed that Smad3 knockout mice were resistant to 12-O-tetradecanoylphorbol-13-acetate (TPA)–induced epidermal hyperproliferation. Concurrently, increased apoptosis was observed in TPA-treated Smad3^−/− skin and papillomas when compared with those of wild-type mice. Expression levels of activator protein-1 family members (c-jun, junB, junD, and c-fos) and transforming growth factor (TGF)-α were significantly lower in TPA-treated Smad3^−/− skin, cultured keratinocytes, and papillomas, as compared with Smad3^+/+ controls. Smad3^−/− papillomas also exhibited reduced leukocyte infiltration, particularly a reduction of tumor-associated macrophage infiltration, in comparison with Smad3^+/+ papillomas. All of these molecular and cellular alterations also occurred to a lesser extent in Smad3^+/− mice as compared with Smad3^+/+ mice, suggesting a Smad3 gene dosage effect. Given that TGF-β1 is a well-documented TPA-responsive gene and also has a potent chemotactic effect on macrophages, our study suggests that Smad3 may be required for TPA-mediated tumor promotion through inducing TGF-β1–responsive genes, which are required for tumor promotion, and through mediating TGF-β1–induced macrophage infiltration.