A group contribution method (GCM) combined with least squares support vector machine (LSSVM) algorithm was used to correlate and predict the speed of sound (SS) of ionic liquids at atmospheric pressure. The NIST Standard Reference Database was used to compile a dataset comprised 41 ionic liquids and consisted of 446 experimental data values. Instead of modelling using a pre-selected and fixed number of functional groups, Forward Feature Selection was combined with the LSSVM algorithm to select the most effective variables, while keeping the number of model parameters as low as possible. The result is an 8-parameter model which has the capability of prediction as well as correlation of the speed of sounds of ionic liquids. The proposed model has an average absolute relative deviation (AARD%) of 0.36%, a coefficient of determination (R-2) of 0.997, and a root mean square error (RSME) of 8.47 ms(-1). (C) 2014 Elsevier B.V. All rights reserved.