The decision to classify a chemical as a human carcinogen must depend upon agreed conclusions from epidemiology, bioassays, and some short-term corroborating tests; information from only one of these disciplines is inadequate. Most pitfalls appear in interpreting the results from animal bioassays; this report will concentrate on them. Often the conclusion is accepted that a chemical is an animal carcinogen without a critical appraisal of the experimental design. By manipulating the experiment, 90 + % of all chemicals can induce some tumor in a rodent. Pitfalls encountered in bioassays result from not specifying the exact agent under test and how it relates to human exposure, using inappropriate routes of administration unrelated to humans, administering illogically high doses, or concluding that a cancer was induced without adequate histopathological description of the lesion. Importance of animal husbandry is often overlooked. Pitfalls are also related to short-term tests. Finally, a major pitfall in assessing carcinogenic risk from chemicals is drawing global conclusions about the carcinogenicity of an agent after the detection of only one or two tumors in the treated group.