Animal models provide unparalleled mechanistic insights into cancer development and potential opportunity for cancer prevention. Nevertheless, species differ markedly with regard to dietary exposures, cancer development, drug effects, and toxicity thresholds; therefore, testing in a single animal system may not predict human responses. Although replication of human cancer in animal models remains inexact, more than two decades of research have clearly yielded significant gains, as is evident in agents tested--and in certain cases, approved--for the prevention of epithelial cancers. Research efficiencies achievable through preliminary testing in genetically engineered and carcinogen-induced animal models enable us to probe genetic and signaling pathways that drive normal and neoplastic processes, and thereby figure prominently in decision trees for agent development.