DNA damage from polycyclic aromatic hydrocarbons (PAH) and other aromatic/hydrophobic compounds has been implicated in case-control studies as a risk factor for lung cancer, as have common polymorphisms in the glutathione S-transferase (GST) genes involved in carcinogen detoxification. However, their joint effects have not been evaluated in prospective studies, leaving open questions about predictive value of these biomarkers. In this matched case-control study nested within the prospective Physicians' Health Study, we evaluated whether biomarkers measured in white blood cells (WBC) significantly predicted risk, alone and in combination, after controlling for level of smoking. The biomarkers reported here are aromatic/hydrophobic-DNA adducts and polymorphisms in genes coding for the GSTM1 and GSTP1 enzymes. Our study population was composed of 89 cases of primary lung cancer and 173 controls, matched in a 1:2 ratio on smoking, age and duration of follow up. Adducts were measured in WBC DNA by the nuclease P1-enhanced (32)P-post-labeling method. Genotypes (GSTM1 null versus non-null and GSTP1 Val versus GSTP1 Ile) were determined by genomic amplification and restriction fragment length polymorphism analysis. Among current smokers, adducts were significant predictors of lung cancer risk (after adjusting for GST genotypes, OR = 3.10, 95% CI 1.07, 9.01). The combined GSTM1 null/GSTP1 Val genotype was associated with lung cancer overall and especially among former smokers, before and after adjusting for adducts (OR for former smokers = 4.21, CI 1.08, 16.41; adjusted OR = 4.68, CI 1.17, 18.71). Among cases only, adducts were significantly higher among current or former smokers with the GSTM1 non-null/GSTP1 Ile genotype. The two risk factors (adducts and genotypes) appear to be independent predictors of risk. The findings underscore the complex and important role of biological susceptibility as a determinant of risk from carcinogens found in tobacco smoke and other environmental compounds.