In this work, GC-sPC-SAFT (Group Contribution Simplified Perturbed Chain Statistical Associating Fluid Theory), and GC-PPC-SAFT (including the polar term related to dipole and quadrupole interactions) equations of state (EoS) were applied in order to correlate thermophysical properties of ionic liquids (ILs), and to predict CO2 and CH4 solubilities in ILs. Previous works from literature proposed an equation in which the parameters for the pure components were estimated from density data, obtaining a correlation of the IL's molecular parameters as a linear function of the molecular weight, and the associative parameters were "transferred" from 1-alkanols. This work proposes four strategies to parameterize ILs from [C-n-mim][BF4], [C-n-mim][PF6] and [C-n-mim][NTf2] families with GC-sPC-SAFT, including the usage of group contribution (GC) methods for ILs, speed of sound in the pure IL parameterization, and reevaluating the association parameters. The model parameters were obtained fitting simultaneously density and speed of sound experimental data at several temperatures, decreasing deviations between experimental and calculated values for these properties with 2B, 3B and 4C schemes. The obtained ILs' parameters were used to predict the solubility of CO2 and CH4 in them. The results obtained in this work with the proposed strategy are able to predict the solubility of gases with high accuracy in a purely predictive way, avoiding the usage of binary interaction parameters, and being compatible with GC methods, and show that both use of speed of sound data and estimation of the association parameters are the key for successful predictions, mainly the use of speed of sound data in the IL parameterization. (C) 2018 Elsevier B.V. All rights reserved.