In order to effectively solve the parameter selection problem of variational mode decomposition (VMD) and accurately extract the bearing fault features, a bearing fault diagnosis method based on multi-feature optimized VMD and fusion index is proposed. Considering the multiple features of fault pulse when the bearing fails, the objective functions and fusion index of information entropy, correlation coefficient, and kurtosis are established, and the parameter optimization problem of VMD is transformed into a multi-objective optimization problem. Firstly, the multi-objective particle swarm optimization (MOPSO) algorithm is used to optimize the three objective functions, and the optimal Pareto frontier solution set of VMD parameter combination is obtained. Secondly, the fusion index is used to evaluate the Pareto frontier solution set, from which the optimal parameter combination of VMD is determined. The bearing fault signal is decomposed by VMD based on the optimal parameter combination, and several intrinsic mode functions (IMFs) are obtained. Then, the fusion index is used to select the optimal IMF, and fault features are extracted. Finally, the analysis results of the simulation signal and actual bearing vibration signals show the effectiveness of the proposed method.