Atmospheric PAH concentrations were determined in Gulbahce district of Bursa, Turkey between August 2004 and April 2005. Measured PAH concentrations were classified as heating and non-heating season samples. The concentrations of total PAHs in heating season were almost ten times higher than those in non-heating season. Diagnostic ratios and factor analysis results show that in the heating season traffic along with residential heating emissions heavily influence PAH concentrations. The plot of logK, versus logP(L)(0) for all the data set of heating and non-heating season samples gave significantly different slopes. The slope for the heating season samples (-0.92) was steeper than the one for the non-heating season samples (-0.78). The partitioning results for individual samples further indicated that slope values varied depending on air parcel trajectories. Air parcels traveled over water (either over the Black Sea or Aegean Sea) prior to arriving at the sampling site had less steep slopes. Partitioning of PAHs was also investigated by comparing experimentally determined K-p values with the results obtained both from octanol-based model (K-p(Oct.)) and soot and octanol-based model (K-p (Soot+Oct.)). Both models were useful in predicting the experimental K-p values. However, they did not explain the observed variability in the experimental K-p values. (c) 2007 Elsevier B.V. All rights reserved.