Polycyclic aromatic hydrocarbons (PAHs) are hazardous air pollutants formed during incomplete combustion, absorbed through inhalation and ingestion, and metabolized to hydroxylated compounds that can be detected in urine. Biomonitoring data provide a direct way to link human exposure to environmental contaminants. However, these data do not reveal how various exposure routes or media contribute to the body burden of a specific chemical. We evaluated predictors of urinary PAH concentrations in 2001–2006 NHANES participants from reported information on demographic and housing characteristics, reported food intake, and modeled outdoor air pollutant exposures. NHANES participants were linked to their daily PM2.5 exposure estimate and annual air toxics concentrations. Multivariate linear regression models were developed using the Deletion/Substitution/Addition algorithm to predict urinary PAHs. Exposure to air pollution was not associated with levels of urinary PAH metabolites. Current smoking status was the strongest predictor of PAH biomarker concentration and was able to explain 10–47% of the variability of PAH biomarker concentrations. In non-smokers, our prediction models were able to explain only 2–5% of the variability of PAH biomarker concentrations. Overall, our results indicated, with the exception of smoking status, there are not strong demographic, dietary, or environmental predictors of PAH exposure.