Exposure to Polycyclic Aromatic Hydrocarbons (PAHs) occurs by respiratory, digestive and dermal absorption. Biomonitoring takes all pathways into account but sensitive and specific biomarkers are required. Different gaseous PAHs metabolites were used due to their abundance in the atmospheric mixtures but none of them were selected as better biomarker than the others. To identify the best candidates for assessing occupational airborne exposure, relation between atmospheric levels of Naphtalene, Fluorene and Phenanthrene and urinary metabolites concentrations was studied in a carbon electrode workers group. Linear mixed effects models were built to select explanatory variables and estimate variance component. Urinary creatinine was a predictor of metabolites levels confirming the importance of diuresis for interpreting results. High significance of pre-shift sampling time combined with positive coefficients of post-shift indicated that urine should be sampled at the end of the workday in association with pre-shift urine to avoid misinterpretations. Among the 10 metabolites studied, urinary 2-hydroxyfluorene and 2-hydroxyphenanthrene showed the highest increase of variance explained by models after inclusion of individual atmospheric levels as explanatory variable. Priority could be given to 2-hydroxyfluorene due to higher excretion levels than 2-hydroxyphenanthrene. (C) 2017 Elsevier B.V. All rights reserved.