Prediction of thermodynamic behavior is essential for the early design stage of separation processes including solvent selection, process optimization and its performance evaluation. In order to better utilize ionic liquids (ILs) as solvents in gas separation processes, the UNIFAC-IL model of IL-liquid solute systems is extended to IL-gas systems by using experimental data from published works and pseudo-experimental data specifically generated from a calibrated COSMO-RS model. In this work, we consider in the model development a total number of 100 ILs from 6 cation families (i.e. imidazolium, pyridinium, pyrrolidinium, ammonium, phosphonium, guanidium) and 24 anion families (i.e. bis(trifluoromethanesulfonyl) amide, tetrafluoroborate, hexafluorophosphate, dimethylphosphate, heptafluorobutyrate, trifluoroacetate, trifluoromethanesulfonate, methylsulfonate, methylsulfate, ethylsulfate, nitrate, p-toluenesulfonate, 2-(2-methoxyethoxy)ethylsulfate, chloride, bromide, dicyanamide, tetracyanoborate, tris(pentafluoroethyl)trifluorophosphate, lactate, levulinate, saccharinate, succinamate, tetrafluoroethanesulfonate, bis(2,4,4-trime-thylpentyl)phosphinate), and 13 gases including CO 2 , SO 2 , H 2 S, NH 3 , N 2 O, CO, N 2 , O 2 , H 2 , CH 4 , C 2 H 4 , C 2 H 6 , C 3 H 8 . The extended UNIFAC-IL-Gas model consists of two sub-models, namely the UNIFAC-IL-Gas (Exp.) model and the UNIFAC-IL-Gas (Pseudo-Exp.) model. The training and testing of the UNIFAC-IL-Gas (Exp.) model is based on 100% experimental data, while the training of the UNIFAC-IL-Gas (Pseudo-Exp.) model is based on pseudo-experimental data, but its testing is also based on 100% experimental data.