In summary, although some of the more optimistic aspirations for human biomonitoring studies envisaged a decade ago have not been realized thus far, some considerable advances have been made. The examples cited above indicate that the feasibility of biomonitoring has been clearly established. In addition, they demonstrate the need for preliminary biomarker testing and validation through transitional studies prior to their field application. In the next decade of research into carcinogen adducts in humans, continued improvements in the reproducibility and specificity of assays for DNA adducts will be needed. Perhaps the increasing use of hybrid methodologies to concentrate adducts followed by specific chemical analyses will allow such adducts to be monitored more precisely. Of course, further basic research into the mechanisms of carcinogenesis will allow the measurement of specific novel markers which are more closely tied to the disease endpoint than adducts. The development of new assays for determining metabolic phenotypes and genotypes relevant to carcinogenesis should improve our estimates of susceptibility (46-48). Such new approaches along with the sustained improvement of current assays will allow molecular approaches to continue to enrich cancer epidemiology in the future.