Effects of dietary supplementation with the antioxidants ellagic acid, quercetin and vanillin were examined using a medium term multi-organ carcinogenesis model in rats. Groups of 10-15 male F344 rats were given i.p. injections of diethylnitrosamine (DEN, 100 mg/kg body wt.) and N-methylnitrosourea (MNU, 20 mg/kg body wt), s.c. injections of 1,2-dimethylhydrazine (DMH, 40 mg/kg body wt.), together with 0.05% N-butyl-N-(4- hydroxybutyl)nitrosamine (BBN) and 0.1% 2,2'-dihydroxy-di-n-propylnitrosamine (DHPN), both in the drinking water, for a total multiple initiation period of 4 weeks (DMBDD) treatment). Ellagic acid, quercetin or vanillin, each at a dose of 1% each in the diet were administered from 1 day before and throughout the carcinogen exposure period, or after completion of the initiation regimen. All surviving animals were sacrificed at the end of week 36, and major organs were examined histopathologically. In the small intestine, significant reductions in the incidence and number of tumors (adenomas and carcinomas) were observed in the groups administered ellagic acid during (8%, 0.08 +/- 0.29) or after (8%, 0.08 +/- 0.29) DMBDD treatment, and those receiving quercetin after DMBDD treatment (0%) compared to the control value (57%, 1.07 +/- 1.21). Although the incidences were not statistically significant, slightly decreased numbers of small intestinal tumors were found in the groups receiving vanillin during (0.33 +/- 0.72), or after (0.40 +/- 0.83) DMBDD treatment. The incidence of large intestinal carcinomas in the group treated with vanillin during DMBDD treatment was significantly higher (73%) than the control value (21%). These results indicated that while ellagic acid and quercetin exerted potent chemopreventive action in both the initiation and promotion stages in the present experimental system, their beneficial effects were restricted to the small intestine. Since small intestinal carcinomas are very infrequent in humans, the advantages of these phenolic compounds for human application as chemopreventors should not be overestimated.