Carcinogen-DNA adducts are generally regarded as relevant biomarkers of carcinogen exposure and their levels in target tissues have often been predictive of tumor incidence in experimental animals. Thus, human risk assessment procedures have utilized dose-response models that assume proportional relationships between carcinogen exposure and cancer susceptibility, even though wide inter-individual variations in human metabolic activating enzymes have now been clearly established. To evaluate these approaches, we have examined the relationship between carcinogen exposure, DNA adduct levels, metabolic activation phenotypes, and cancers of the larynx, urinary bladder, and colon. Cigarette smoking is a strong risk factor for cancers of the larynx and urinary bladder. In the larynx, the DNA adducts appear to be derived predominantly from polycyclic aromatic hydrocarbons (PAHs) and are evident only in tissue from smokers. However, adduct levels appear to be determined primarily by expression of cytochrome P450 (CYP) 2C9/10, which varies > 10-fold in different individuals. This CYP catalyzes the metabolic activation of benzo (alpha) pyrene (BP) to a 9-hydroxy-BP-DNA adduct that accounts for up to 25% of the putative PAH adducts formed in vivo. For the urinary bladder, putative aromatic amine (AA)-DNA adducts are predominant and are significantly elevated in current smokers. Rapid CYP1A2 and slow acetyltransferase (NAT2) phenotypes have been previously implicated in the activation (N-oxidation) and detoxification (N-acetylation) of AAs for human bladder carcinogenesis. Data now indicate that NAT1, which is expressed in human urothelium and catalyzes the O-acetylation of N-hydroxy arylamines, is significantly correlated with DNA adduct levels and is bimodally distributed in this tissue. Colo-rectal cancer risk, which has been associated with exposure to heterocyclic amines (HAs) in cooked foods, is strongly elevated in individuals with the combined rapid phenotypes for CYP1A2 and NAT2. These enzymes are uniquely responsible for HA N-oxidation and subsequent O-acetylation, forming DNA adducts that are found in human colon. These studies indicate that cancer risk assessment procedures should be redesigned to include biomarkers of susceptibility, especially those involved in carcinogen bioactivation.