It is well-known in reversed-phase liquid chromatography that the solute retention factor (k) in binary eluents is well modeled as a quasi-linear function of the eluent composition (Phi) by the equation log k = log k(s/w) - SPhi, where the solute parameters k(s/w) and S are the retention factor in pure water and the solvent strength parameter, respectively, and Phi is the volume fraction of the organic component of the eluent. S is related to the free energy of the solute transfer from water to bulk organic liquid, which in turn is related to the infinite-dilution activity coefficients (gamma(infinity)) of the solute in water and bulk organic liquid. We report that the gamma(infinity) values obtained from a gamma(infinity) prediction model, UNIFAC, can be used to predict S directly and also to aid in the determination of best-fit k(s/w) values for acetonitrile-water mobile-phase systems. More specifically, UNIFAC-based S values combined with experimentally determined k(s/w) values for a limited set of solutes can be used to predict the variations in retention as a function of mobile-phase composition in acetonitrile-water systems within the calibration range for solutes that were not in the original data set.