As the molecular basis of carcinogenesis becomes understood, monitoring key genetic alterations associated with carcinogenesis will play an ever-increasing role in toxicological evaluation. Also, molecular biological techniques can be used to construct animal models that include key molecular features of the human carcinogenesis process. Such approaches to carcinogenesis evaluation have already begun to be used in regulatory practice, and they can be expected to play a more prominent role in the future. For example, the International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use recently adopted testing guidelines that allow mechanistic short-term models of carcinogenesis to be incorporated into regulatory submission for approval of new pharmaceuticals (DeGeorge, 1998). EPA draft carcinogenicity evaluation guidelines that encourage incorporation of mechanistic data (i.e., related to mode of action) into cancer risk assessments are currently pending adoption (Federal Register, 2003). Mechanistically based animal carcinogenicity models currently under evaluation currently include the rasH2 mouse model that incorporates the normal human ras protooncogene into a mouse model, the p53 mouse model, based on the recognition of the importance in the human of changes in p53 function for expression of carcinogenesis, and several other mechanistically based models including some with inactivated defense or repair genes (e.g., Robinson and MacDonald, 2001). Thus, the tools of genetics are already being employed to construct appropriate "humanized" genetic models for carcinogenicity testing and also to monitor genetic changes associated with human and animal carcinogenesis as a part of product development. As the specific processes involved in human carcinogenesis, including the role of defense and repair systems, become better understood, it can be expected that animal models will be selected or constructed that allow more accurate and earlier prediction of induction of events likely to result in human cancers. Differences in these factors between humans and rodents are already becoming understood for a limited number of well-studied cases such as UV-induced mutation and cancer (Tang et al., 2000). This is likely to make feasible the monitoring in animal models those specific genetic events associated with human cancer, and also may permit monitoring of those same genetic events directly in the human.