Urinary hydroxylated metabolites of polycyclic aromatic hydrocarbons (PAH) were investigated as potential biomarkers of bitumen exposure in a cross-shift study in 317 exposed and 117 non-exposed workers. Personal measurements of the airborne concentration of vapours and aerosols of bitumen during a working shift were weakly associated with post-shift concentrations of 1-hydroxypyrene (1-OHP) and 1-, 2+9-, 3- and 4-hydroxyphenanthrenes (further referred to their sum as OHPHE), but not 1- and 2-hydroxynaphthalene (OHNA). Smoking showed a strong influence on the metabolite concentrations, in particular on OHNA. Pre-shift concentrations of 1-OHP and OHPHE did not differ between the study groups (P = 0.16 and P = 0.89, respectively). During shift, PAH metabolite concentrations increased in exposed workers and non-exposed smokers. Statistical modelling of post-shift concentrations revealed a small increase in 1-OHP by a factor of 1.02 per 1 mg/m(3) bitumen (P = 0.02) and 1.04 for OHPHE (P < 0.001). A group difference was observed that was diminished in non-smokers. Exposed non-smokers had a median post-shift 1-OHP concentration of 0.42 μg/l, and non-smoking referents 0.13 μg/l. Although post-shift concentrations of 1-OHP and OHPHE were slightly higher than those in the general population, they were much lower than in coke-oven workers. The small content of PAHs in vapours and aerosols of bitumen, the increasing use of additives to asphalt mixtures, the strong impact of smoking and their weak association with airborne bitumen limit the use of PAH metabolites as specific biomarkers of bitumen exposure.