Cancer prevention has been the stated goal of molecular cancer epidemiology for the past 17 years. In this review, progress toward that goal is evaluated by using as examples well-studied environmental exposures-i.e., tobacco smoke, polycyclic aromatic hydrocarbons, aflatoxin B(1), benzene, and hepatitis B virus-and their roles in lung, breast, and liver cancers and leukemia. The contributions of molecular epidemiology discussed here include providing evidence that environmental agents pose carcinogenic risks, helping establish the causal roles of environmental factors in cancer, identifying environment-susceptibility interactions and populations at greatest risk, and developing new intervention strategies. Molecular epidemiologic and other data indicate that assessment of carcinogenic risks should address both the range of risk across the population and the risk to subgroups who may be at high risk because of genetic or acquired susceptibilities, including young children. However, for the most part, research results have not yet been effectively translated into risk assessments and preventive health policies. An infrastructure linking scientists, policy makers, and other constituencies is needed to facilitate this process. To extend our knowledge, the second generation of molecular epidemiologic research should include large-scale, collaborative studies incorporating validated biomarkers and automated technologies. An incentive to make the necessary investment is the recognition that prevention of only 20% of cancer in the United States would result in 200000 fewer new cases diagnosed each year and an annual savings of $21.4 billion in direct costs alone.