Animal models for the study of toxicity are chosen more for pragmatic reasons such as life‐span, ease of handling, and economics than because of a comprehensive process of validation. There is also a tendency to seek the answers to nonspecific questions and then to use those answers in situations where they cannot lead to proper judgments. In some cases, such as in carcinogenicity testing, an uneasy compromise exists between the sensitivity of the model and the background noise. There is scope, by increased attention to both the control of environmental factors and the genetics of the animal models, to increase the sensitivity and validity of the systems that are used in toxicology.