Abstract: Although pre-stack inversion technology can be used to obtain multiple fluid identification factors, multi-solutions are generally caused by reservoir prediction using a single fluid identification factor. At present, the comprehensive interpretation of reservoir based on various fluid identification factors has become a new trend, of which most methods rely on experts and their experience. Therefore, the random forest algorithm is introduced into reservoir fluid identification. Firstly, the input features (fluid identification factors) are selected based on logging data, and the effects of the number of input features and different feature combinations on the prediction results are studied. Then the algorithm is used to learn the nonlinear relationship between the input features and the reservoir information of the well. Finally, according to the obtained results, a comprehensive analysis is performed on the reservoir, thus realizing the integrated utilization of a large number of fluid identification factors. The algorithm weakens the multi-solution caused by a single fluid identification factor, and improves the accuracy and reliability of reservoir fluid identification. The application examples show that five kinds of fluid identification factors and reservoir information in wells are comprehensively explored using random forest algorithm, achieving the accurate identification of gas and water in underground reservoirs.