The importance of inherited mutations as a cause of human disease has been established clearly through examples of well-defined genetic anomalies, such as Down syndrome and retinoblastoma. Furthermore, it is suspected that environmental contaminants induce mutations resulting in increased risk for such defects in subsequent generations of persons exposed. The present lack of direct evidence for induced inherited genetic disorders in human beings hampers the development of risk estimation techniques for extrapolation from animal models. The most extensive prospective epidemiologic studies of inherited genetic effects have involved survivors of atomic bomb detonations and patients treated with cancer chemotherapy. In neither case has a significant elevation in inherited genetic effects or cancer been detected in the offspring of exposed individuals. Epidemiologic studies of subjects receiving chronic exposure may be confounded by the effect of maternal exposure during pregnancy. Consideration of only paternal exposure can minimize the confounding influence of teratogenicity, enhancing the resolving power of studies for inherited effects. Using this approach, retrospective (case-control) studies of childhood cancer patients have provided limited but suggestive evidence for inheritance of induced effects. Endpoints, such as congenital malformations and spontaneous abortion following paternal exposure, can also be considered as indicators of heritable mutagenic effects. For example, there is limited evidence suggesting that paternal exposure to anaesthetic gases may cause miscarriage and congenital abnormalities as a result of induced male germ cell mutations. By comparing male-exposure endpoints for which there are human data, as described above, with parallel or similar animal endpoints, such as dominant lethal, inherited cancer and "male teratogenic" effects, it is possible that suitable models for extrapolating to human risk can be developed. In order to establish a clearer relationship between induced mutation and genetic disease, the current surveillance systems should be expanded to include endpoints relevant to genetic study. The relaxation of regulations regarding access to census data could improve the chances of documenting such an association.