Successful efforts have been made at predicting the adhesion between several polymer matrices and the surface of a glass filler, treated with various coupling agents using thermodynamic descriptors. A good thermodynamic criterion has been shown to be the (negative) Gibbs free energy of mixing (-G(mix))(0.5), for a solution containing molecules representative of the repeat units of the polymer and the exposed functional groups of the silane coupling agents. Group contribution methods, such as UNIFAC allow for straightforward calculations of (-G(mix))(0.5); however, the UNIFAC database remains incomplete, and furthermore requires a subscription to access the most current database. Recently, new predictive models based on the surface charge distribution of molecules and statistical thermodynamics have been developed to predict properties of solutions. In this work, one such method, COSMO-SAC, is used to calculate (-G(mix))(0.5) and compare the results to a previously determined database of relative practical adhesion strengths between various polymer matrices and the organo-functional groups of silane coupling agents. The results obtained from COSMO-SAC show good agreement with the experimental database, except for the cases of significant steric effects and systems containing amines. Overall, the adhesion predictions obtained from COSMO-SAC calculations are comparable to the UNIFAC predictions, although UNIFAC was more successful in predicting adhesion for the amine containing compounds.