Abstract: In order to improve the prediction accuracy on internal corrosion rate of multiphase flow pipelines, the combined model of Improved Gravitational Search Algorithm (IGSA) and Random Forest Regression (RFR) was adopted, the number of decision trees and division characteristics of RFR model were optimized by IGSA, the influence of iteration times on prediction accuracy was discussed, the accuracy of model was verified by introducing the statistical indicators, and the characteristic variables affecting the corrosion rate were ranked. The results show that the variables other than pressure have great influence on the internal corrosion rate. The mean-square error of IGSA-RFR combined model in the same test set is 0.000 016 5, which is 2 orders of magnitude smaller than that of other models. Besides, its mean absolute percentage error is 0.524 1, which is 1 order of magnitude smaller than that of other models, and the coefficient of determination R² is 0.999 6, which is the closest to 1 in all models. Hence, IGSA-RFR model is better than other models in terms of prediction accuracy and suitable for the research on prediction of internal corrosion rate of multiphase flow gathering pipelines.