A trained version of Jouyban-Acree model is proposed employing 32 solubility data sets of 5 drugs in ethanol + propylene glycol mixtures at various temperatures. Using this model, the solubilities of a drug in the mono-solvents and the Abraham solvation parameters are required to predict the solubility in the binary solvent mixtures. The overall mean percentage deviation for the correlated data was 11.0 %, and that of a predicted data set was 11.2 %.