Recently several publications appeared correlating octanol-water partitioning coefficients (KO/W) with solid-phase microextraction (SPME) extraction coefficients on poly(dimethylsiloxane) (PDMS) fibers. This correlation seems very good for medium-polar to polar compounds but cannot explain the observations for apolar compounds. It is shown that for polychlorinated biphenyls (PCBs) the published data are erroneous, because of system adsorption effects. PCB concentrations up to 10 times higher were measured on the stir bar compared to the SPME fiber. Using a short packed PDMS trap, it is shown that the true PDMS-water equilibrium constant is indeed proportional to literature KO/W data.