This data in the aggregate suggests that the 3 best studied genetic susceptibility factors (CYP2D6 extensive metabolizers, GST mu null phenotype, and CYP1A1 "mutant" alleles in Asians only) constitute greater risk factors for the more smoking related histologies of lung cancer, but not for adenocarcinoma. The epidemiologic evidence for a these genetic susceptibility factors in tobacco-related cancer is suggestive but not determinant. A consensus estimate of relative risk for extensive metabolizers of debrisoquine is around 2. Variability in study results depend on a number of factors which include: assay misclassification, non-correspondence of phenotype/genotype in certain subjects, disease heterogeneity, exposure variation, ethnic and racial variation. Future studies should emphasize: a high quality approach to data gathering, careful attention to epidemiologic design, and the use of intermediate markers where feasible. Investigators should consider the use of multiple genetic markers since PCR approaches can make this an efficient approach. A meta-analysis may serve to illuminate points of heterogeneity between studies. New discoveries should provide opportunities to explore for analogous associations in other malignancies. It may be speculated that the "specificity" of the association observed for each of the genetic factors tends to support the general causal nature of the hypothesis. The fact that each shares the histologic preference at least suggests that a common mechanism may be operative. The observation that the tobacco-cancer association, though clearly present, is weaker for adenocarcinoma than for the other lung cancer histologies, suggests that the underlying mechanism involves some interaction of the genetic trait with exposure to tobacco smoking, and suggests further attention to this factor to elucidate differences in risk estimates for genetic susceptibility factors among different studies.