Prediction of acid gases solubilities in ionic liquids (ILs), have recently emerged as promising mediums for refining of natural gas, using powerful paradigms is of great importance from technical and economical point of view. In this respect, this study aims at appraising the effectiveness of one of the new generation soft computing methodologies called gene-expression programming (GEP) for estimating the hydrogen sulfide (H2S) solubility in ionic liquids (ILs). A total data set of 465 experimental data belonging to 11 ionic liquids, which gathered from literatures, were used to develop a general correlation. The temperature and pressure accompanied with acentric factors and critical temperature and pressure of ILs were used as independent input variables, while H2S solubility as dependent output variables. The modeling results showed the coefficient of determination (R-2) of 0.9902 and 0.0438% mean absolute relative error (MARE) for the predicted solubilities from the corresponding experimental values. Therefore, the model is comprehensive and accurate enough to be used to predict the H2S solubility in various ILs. In addition, the GEP-model predictions were compared with the outputs of two well-known engineering approaches named Soave-Redlich-Kwong (SRK) and Peng-Robinson (PR). Results showed that the proposed evolutionary-based method was more accurate than the widely used aforementioned thermodynamic models. (C) 2014 Elsevier B.V. All rights reserved.