Criteria for definition of a human teratogen are similar to Koch's postulates and include (1) presence of the agent during the critical period of development, (2) production of congenital defects by the agent in an experimental animal, and (3) evidence that the agent acts directly on the embryo or fetus. Examples of how teratogens have been identified through specific syndrome identification and by application of experimental animal information are given. Hyperthermia, an example of the latter, is discussed in some detail. Certain bias factors that confound epidemiologic studies are discussed. A system for detection an prevention of human teratogenicity is proposed. The system is based on the integration of three data bases: (1) the exposed parent-child pair, (2) the chemical and biologic effects of the agent, and (3) identification of specific exposure syndromes.