The effects of cancer predisposition and increased tumorigenic radiosensitivity of the predisposed genotypes on radiation cancer risks (in the general population and in sisters and first cousins of affected probands) are studied using an autosomal dominant model of cancer predisposition and radiosensitivity. The model assumes that the predisposing alleles, which confer enhanced tumorigenic radiosensitivity, are incompletely penetrant. In addition, the model also allows for sporadic cancers, unrelated to the predisposing locus. The predictions of the model are illustrated using current estimates of BRCA1 mutant gene frequencies; the estimates of the strength of predisposition and radiosensitivity differentials used are based on animal and human studies. It is shown that, unless both the strength of predisposition and radiosensitivity differential are large (say, > 100-fold in comparison with normal homozygotes), (i) the effect of risk heterogeneity on cancer risk is marginal; (ii) dose-dependent radiation effect remains virtually the same as in a homogeneous irradiated population that has no predisposed subgroups; (iii) for the same radiation dose, relatives of affected probands show an enhancement of cancer risks; and (iv) most extra cancers in relatives can be attributed to radiosensitivity differentials. This simple model can give an upper bound of the effect of risk heterogeneity on radiation-induced breast cancer risks even when the cumulative breast cancer risk is age-dependent. Further, our model predicts that the benefits of mammography outweigh the risks.