LAUSR

The ability of a number of common models for correlating the collected solubility data of drugs in ethanol + water mixtures (the number of data sets is 43) was tested. The selected models are classified within three groups, (1) λ–h, van’t Hoff and Apelblat models with temperature dependency, and modified Wilson with solvent composition dependency, (2) Jouyban–Acree, Jouyban–Acree–van’t Hoff and Jouyban–Acree–Apelblat (linear models) and NRTL, Wilson and UNIQUAC (nonlinear models) with both temperature and solvent composition dependencies, (3) UNIFAC, Yalkowsky, log–linear model of Yalkowsky, Jouyban–Acree–Abraham and Jouyban–Acree–Hansen predictive models. Of the models in group 1, the Apelblat model has the lowest overall %ARD and Jouyban–Acree–van’t Hoff and Jouyban–Acree–Apelblat models in group 2 performed best in correlations. Moreover, predictions using Jouyban–Acree–Abraham and Jouyban–Acree–Hansen models in group 3 have the lowest %ARDs in this group.