A preferential solvation model that relates solute properties with solvent composition in binary mixtures has been applied to the dissociation pK a values of a set of 28 substituted phenols in methanol-water mixtures. The parameters of the model allow estimation of the pK a value of each phenol for any methanol-water composition. Moreover, it is demonstrated that the pK a values of the whole set of phenols at any methanol-water composition are linearly related to the pK a values of the phenols in water. Equations that relate the correlations' slope and intercept values with the solvent composition have been derived and tested with the set of phenols. The general parameters obtained for these equations allow an accurate calculation of the pK a value of any phenol, even of those not included in the original set, at any methanol-water composition solely from the pK a value of the phenol in water. These calculated pK a values can be used for quantitative structure-HPLC retention relationships. The method is tested by comparison of the calculated pK a values with the HPLC determined pK a values of 26 phenols in a polymeric column with a 50% methanol as mobile phase.