Retinoic acid is one of the most promising drugs for chemotherapy and chemoprevention of cancer. Either blocking activator protein-1 (AP-1) activity or activating retinoic acid response element (RARE) have been proposed to be responsible for its antitumor activity. However, evidence for this hypothesis is lacking in vivo studies. To address this issue, we used an AP-1-luciferase transgenic mouse as a carcinogenesis model and new synthetic retinoids that are either selective inhibitors of AP-1 activation or selective activators of the RARE. The results showed that the SR11302, an AP-1 inhibition-specific retinoid, and other AP-1 inhibitors such as trans-retinoic acid and fluocinolone acetonide, markedly inhibit both 12-O-tetradecanoylphorbol-13-acetate-induced papilloma formation and AP-1 activation in 7,12-dimethyl benz(a)anthracene-initiated mouse skin (P < 0.05). In contrast, repeated applications of SR11235, a retinoid with RARE transactivating activity, but devoid of AP-1 inhibiting effect, did not cause significant inhibition of papilloma formation and AP-1 activation (P > 0.05). These results provide the first in vivo evidence that the antitumor effect of retinoids is mediated by blocking AP-1 activity, but not by activation of RARE.