In this communication, a hybrid model based on Least Square Support Vector Machine (LSSVM) was constructed to predict CO2 corrosion rate. The input parameters of the model are temperature, CO2 partial pressure, flow velocity and pH. The data used for training and testing of the developed model are 612 and 109 data, respectively. In order to benefit LSSVM from Kernel learning, we compared three kernel functions to select the most efficient one. Furthermore, Coupled Simulated Annealing (CSA) optimization technique was adapted to choose the best optimal values of the model parameters. The results elucidate that Gaussian Kernel functions is the desired function which can afford high accuracy for predicting CO2 corrosion in oil and gas industries. (C) 2016 Elsevier B.V. All rights reserved.