The vapour-liquid equilibria of ethanol/l-propanol/water mixture saturated with sodium chloride, potassium chloride and copper sulphate predicted by Tan-Wilson and Tan-NRTL phase models compared well with the experimental data. The prediction was based on the solvent-solvent interaction parameters obtained from the regression of the experimental vapour-liquid equilibrium of the salt-free mixture and the solute-solvent interaction parameters calculated from the bubble points of the individual solvent components saturated with the respective salt. The extent of salfing-out of the solvent components from the liquid phase by the three salts generally decreased in the order of 1-propanol, ethanol and water giving a vapour phase rich in I-propanol and ethanol and a liquid phase rich in water at equilibrium. These effects are slightly different between sodium and potassium chloride and are significantly greater than those of copper sulphate. These observed salting effects are consistent with the criteria that solvent component, i, would be preferentially salted-in or out of the liquid phase than solvent component, j, if the ratio of their solute-solvent interaction parameters A(sj)/A(si) (Tan-Wilson) or exp(pi(is) -tau(js)) (Tan-NRTL) is less or greater than one. These results confirm the good accuracy of the two phase models in describing the effect of a non-volatile solute on the liquid phase activity of a solvent component in a ternary solvent mixture. These results suggest that the rapid screening method and criteria for selecting a suitable non-volatile solute to separate close-boiling and azeotropic binary solvent mixture developed by Tan based on these phase models can be extended to include the ternary solvent mixtures. (c) 2005 Elsevier B.V. All rights reserved.