Extrapolation of results from rodent bioassays involving high-dose exposures to possible carcinogenic risk in humans exposed to low doses is based on the assumptions of species relevance and high- to low-dose extrapolation. For genotoxic chemicals, such as 2-acetylaminofluorene and N-[4-(5-nitro-2-furyl)-2-thiazolyl]formamide, these assumptions appear to be appropriate, although the dose response can be greatly modified by cell proliferation effects of these chemicals at high doses. In contrast, nongenotoxic chemicals, such as chemicals causing urinary calculi or sodium saccharin and related sodium and potassium salts, frequently are carcinogenic only at high doses and/or only in specific species. Consequently, for extrapolation of results for nongenotoxic chemicals these assumptions may not be appropriate.