We conducted a cross-sectional molecular epidemiological study of coke oven workers exposed to the established carcinogen polycyclic aromatic hydrocarbons (PAHs) to evaluate the relationships between both traditional 'exposure markers' and a series of biomarkers, including urinary 1-hydroxypyrene as a marker of internal dose, leukocyte aromatic DNA adducts as markers of biologically effective dose, serum p53 protein as a response marker and genetic polymorphisms of cytochrome P4501A1 and glutathione S-transferase MI as susceptibility markers. Twenty-five male subjects each were randomly selected from the top, middle and bottom work areas of the oven, and the control plant. They were matched for age and smoking status. The mean levels of PAH exposure, monitored by stationary and personal samplers, and of worker urinary 1-hydroxypyrene differed significantly between the top, middle and bottom of the oven and control work areas. The highest stationary and personal PAH concentrations and 1-hydroxypyrene levels were demonstrated at the top work area. Good correlations were found between the stationary PAH levels, personal PAH levels and urinary 1-hydroxypyrene levels. No positive correlations were demonstrated between aromatic DNA adduct levels and current or cumulative PAH exposure dose. In the presence of genetic polymorphisms of cytochrome P4501A1, a positive correlation was demonstrated between aromatic DNA adducts and urinary 1-hydroxypyrene levels. There was also a significant correlation between serum p53 protein levels and the cumulated benzo[a]pyrene exposure dose. Although these biomarkers have certain limitations, they are applicable to cancer epidemiology, and may contribute to our understanding of the mechanisms of carcinogenesis.