Emerging short-term bioassays for chemically-induced carcinogenesis are dependent for their relevance to human risk assessment on the degree of coincidence of human and rodent tumor pathways. Since these pathways do not always converge, these new tests may have a number of unanticipated pitfalls. Models of liver and renal tumors are described. The results from Rb and p53 tumor suppressor gene transgenic animals are compared to human tumor syndromes. The question of mutagenic and epigenetic fingerprints of chemicals versus the cell-specific selection of spontaneous mutations is debated. Examples of specific pitfalls, such as the recently discovered Helicobacter hepaticus promoted liver tumors in mice are presented. The rat pseudogenes for p53 and the rare role of p53 in most important rodent tumor models other than epithelial tumors present experimental quandaries. The differential effects of carcinogens during various stages of rodent perinatal and adult development are also discussed. It is concluded that the pathways of both animal models and their human counterparts should be better identified so that realistic endpoint markers can be chosen for human carcinogenic risk assessment.