The relationship between chemically induced patterns of tumorigenesis in rodents and of in vitro genetic toxicity was evaluated for 73 substances. tumorigenicity patterns were defined according to sex and species effects, the induction of common or uncommon tumors, and benign or malignant tumors. These results and the genetic toxicity results derived from the testing of chemicals under code were compared. Chemicals that induced tumors in both sexes of both rodent species (trans-sex/species carcinogens) were divided into those that showed multiple responses in genetic toxicity assays and those that showed little or no response. Some of the nongenotoxic trans-sex/species carcinogens exhibit properties that do not necessarily fit classification as only tumor promoters and may involve some other mode(s) of action. Those chemicals showing tumorigenicity in only one of the four groups exposed (uni-sex/species carcinogens) generally showed little or no response in genetic toxicity assays. Uni-sex/species carcinogens may be difficult to identify by in vitro assays because of their high tissue specificity. Chemicals that are tumorigenic in both sexes of both species are logically more likely to be tumorigenic in a third species than are those that are tumorigenic in only one sex of the exposed species. Therefore, while positive genetic toxicity test results are not predictive of all carcinogens, a consistent positive response among the multiple endpoints in these assays is more likely to identify chemicals with the potential for trans-sex/species carcinogenesis. Such trans-sex/species carcinogens may have the most direct implication for human health effects.