This work is a continuation and extension of previously published study showing that the perturbed-chain statistical associating fluid theory (PC-SAFT) is an efficient and robust approach for calculating thermodynamic properties of the mixtures containing ionic liquids (ILs) [J. Phys. Chem. B 2012, 116, 50025018]. The modeling strategy presented in the original paper was based on application of infinite dilution activity coefficients of molecular solutes in ILs (gamma(infinity)) to calculate binary corrections for the PC-SAFT combining rules. In this work, the idea is further investigated, in particular extended to a broader spectrum of pure fluids (the model parametrization is provided for almost 100 ILs), and evaluated based on a large compilation of liquidliquid equilibrium (LLE) data, including more than 400 distinct and diverse ternary systems and a number of higher systems. Three PC-SAFT modeling strategies are proposed and systematically reviewed in order to elucidate an effect of gamma(infinity) data on the quality of LLE predictions. Experimental versus predicted impact of ILs cation and anion on LLE for different kinds of binary subsystems of molecular compounds is also highlighted and discussed. Finally, the results obtained with the PC-SAFT are confronted with the calculations carried out with conductor-like screening model for real solvents (COSMO-RS), often perceived or referred to as the state-of-the-art predictive tool of modern thermodynamics of liquid solutions.