The rainbow trout has been utilized as a model for human carcinogenesis for a number of years. Trout are relatively inexpensive to maintain and exhibit (over the 9-12-month tumor assay period) very low spontaneous tumor backgrounds. One of the most powerful applications of this model is the design and conduct of large-scale tumor studies requiring thousands of animals that address statistically challenging questions of dose-response. Two recent examples of such applications include our studies on I3C as a tumor promoter and DBP as a tumor initiator. I3C was shown to promote AFB1-initiated liver cancer at doses near those recommended for supplementation in humans. Further studies are required to determine if the mechanisms responsible for promotion in trout can be extrapolated to humans. In the second example, we report results from the largest animal tumor study ever conducted. A total of 42,000 trout were utilized to measure DBP carcinogenesis down to incidences of 1 in 5,000. The dose response model deviated significantly from linearity although the existence of a threshold could not be statistically established. Extrapolation of the data model predicts a DBP dose producing 1 in 10(6) cancers that is 1,000-fold higher than predicted by the conservative linear model. If these results can be confirmed with other carcinogens (genotoxic and perhaps nongenotoxic) and other targets, this could have a significant impact on the utilization of animal tumor data in human risk assessment.