Methods for estimating the risks, that is, the probabilities of contracting a disease, are required as a basis of decision-making regarding the needs for protection and risk reduction. A mechanism-based model has been developed for estimating the cancer risks from genotoxic chemicals using adducts to macromolecules for determining the in vivo dose. On the assumption that cancer is caused by an increased frequency of mutations in tissues, interacting with inherited or acquired growth-promoting factors, a simplified model has been proposed for estimating cancer risks from chemically reactive, that is, mutagenic agents. According to a multiplicative model, the risk increment (deltaP) is deltaP = beta x D x P0, proportional to the background incidence (P0) and linearly dependent on dose at low to intermediate doses (D); beta is the risk coefficient, which is approximately the same for different tumor sites and, probably, different species. This model is already in use for ionizing radiation and has been validated for a few mutagenic and carcinogenic chemical compounds. Inherent in this method is the measurement of dose. Sensitive chemical methods have been developed for determining reactive compounds and intermediates in vivo through their stable reaction products (adducts) with proteins, particularly hemoglobin. In humans or animals, the doses of genotoxic agents can be derived from measured levels of adducts and rates of adduct formation. This approach has been applied to various exposures such as air pollutants in occupational settings, carcinogens in foods, and tobacco smoke. By such methods, exposures to previously unknown mutagens and carcinogens may be detected and assessed in humans. Examples of this are epoxides (from endogenously produced alkenes) and compounds (such as acrylamide) formed in cooking foodstuffs.