A computer implemented software method enables the prediction of the pK.sub.a of an arbitrary molecule based upon a knowledge of the molecular structure of that molecule and a statistical analysis of the molecular structures of a group of molecules (training set) for which the pK.sub.a
