The problem of extrapolating from observed carcinogenic effects to estimates of risk for exposed populations.

A review is presented of the complex problems involved in the quantitative estimation of cancer risk for populations exposed to carcinogens on the basis of evidence provided either by epidemiologic studies or by experimental data. Strain and species differences among experimental animals and wide interindividual differences among human subjects are discussed. Examples from quantitative laboratory studies on carcinogenic risk factors are discussed to illustrate research approaches in this field, especially in measurements of carcinogen interactions with animal and human target tissues, in human tissue-mediated mutagenesis of mammalian cells, and in studies of combined effects of different carcinogens. The scientific criteria for the quantitative estimation of carcinogenic risk, developed by the Interagency Regulatory Liaison Group, are reviewed, and related research needs are indicated. The implications for regulatory policy, as outlined by the U.S. Regulatory Council, are discussed. Precise and reliable quantitative estimates of cancer risk often appear not to be obtainable and, in such cases, nonquantitative risk assessments have been recognized as appropriate.