The preventive efficacy of the triterpenoid 5MeCDDO was tested in two models of mammary cancer, the Min model of intestinal cancer, and a chemically induced model of head and neck cancer. In one model ofmammary cancer, female Sprague- Dawley rats were administered MNU at 50 days of age, and 5MeCDDO (27 ppm) was administered in the diet beginning 5 days later for the duration of the study; 5MeCDDO was ineffective. In contrast, in a model examining initiation of mammary cancers by the procarcinogen dimethyl-benzanthracene, 5, 6-benzoflavone (500 ppm, an Ah receptor agonist) or 5MeCDDO (27 or 2.7 ppm) decreased tumor multiplicity by 90%, 80%, and 50%, respectively. This anti-initiating effect which is presumably mediated by altered metabolic activation parallels our observation that 5MeCDDO induced proteins of various antioxidant response element (ARE)-related phase II drug-metabolizing enzymes [e.g., GST Pi, AKR 7A3 (aflatoxicol), epoxide hydrolase, and quinone reductase] in the liver. 5MeCDDO tested in the 4-nitroquinoline-l-oxide (4-NQO) head and neck cancer model failed to decrease tumor incidence or invasiveness. In the Min mouse model of intestinal cancer, a high dose of 5MeCDDO (80 ppm) was weakly effective in reducing adenoma multiplicity [30% (P < 0.05)]; however, a lower dose was totally ineffective. These findings question whether measuring increased levels of certain ARErelated genes (e.g., quinone reductase, GST Pi), indicating decreased carcinogen activation are sufficient to imply general chemopreventive efficacy of a given agent or mixture.