Genetically engineered mice that either over-express a foreign gene (transgenic) or in which the activity of a specific gene has been removed ("knock-out") or replaced ("knock-in") will be used increasingly to investigate molecular mechanisms of disease, to evaluate innovative therapeutic targets, and to screen novel agents for efficacy and/or toxicity. Recent innovations of relevance to toxicologic researchers include the construction of genetically engineered mice with (1) multiple engineered genes, (2) mutations that can be induced at specific sites and times throughout life, and (3) the substitution of human genes for their mouse counterparts ("humanized" mice) to allow in vivo investigation of xenobiotic toxicity. Contemporary applications of genetically engineered mice in toxicology include basic mechanistic research exploiting newly engineered mouse lines as well as applied screening for genotoxicity and carcinogenicity using commercially available animals. Many caveats must be considered when interpreting genetically engineered mice-derived toxicity data, the chief of which will be the extent to which the model's phenotype has been fully characterized, the type and incidence of background lesions for the given mouse strain and engineered gene, and the possibility of misinterpreting the presence or absence of a phenotype due to compensatory physiologic processes that mask the outcome produced by the engineering event. Toxicity data acquired using genetically engineered mice currently supplements and in time likely will supplant those gathered using the present "gold standard" bioassays, as genetically engineered mice typically develop more lesions after a shorter latency period than do age- and strain-matched, wild-type mice.