This invention relates to a process for using computerized molecular interaction modeling to predict the adhesive interactions between a substrate and a polymer. The molecular modeling method may be used to predict and select optimal adhesion promoting monomers for use in latex polymer coatings, providing the best wet adhesion to alkyd-coated substrates. The molecular modeling method could also predict substrate polymer pairs having the least affinity, and thus the most useful as a release liner. The method involves the steps of: a) identifying interacting chemical segments on both the surface and the polymer; b) generating models of the interacting segments, said models describing the spatial relationship of each atom in the segment and the connectivity between the atoms; c) merging the models of each surface segment with each polymer segment to describe each possible interacting surface/polymer pair; d) generating several hundred random configurations for each surface/polymer pair merged models, by choosing random values for the six spatial variables, that describe the relative orientations of two objects; e) optimizing the atomic coordinates of each surface/polymer segment interaction model by calculating the minimum of the molecular potential energy; f) computing the pair interaction energy for each merged model pair; g) averaging the pair interaction energies; and h) comparing the average pair interaction energies of each surface/polymer pair to choose the best pair for the intended application.