'Single' genes, necessary and sufficient to cause cancer, are contrasted with 'susceptibility' genes that are neither, but may act in the presence of specific environmental exposures to alter the chances of cancer in the host. The former are rare, are of high absolute and relative risks, have minimal dependence on exposures and therefore have low population attributable risks. A small number of such genes are well established in the literature and typically exhibit familial aggregations of disease that serve as a starting point for genetic studies. 'Susceptibility' genes, as typified by the pharmacogenetic model, are common, have low relative and absolute risk, are strongly dependent upon exposure, and may have potentially high population attributable risks. Mechanistic and epidemiologic data are suggestive but currently fall short of confirmation for these associations. Familial aggregation is not a prominent feature and epidemiological study designs with careful exposure assessment is the investigative method of choice. Both approaches require interdisciplinary expertise and benefit from advances in molecular biology.