Abstract: Traditional diagnostic methods for identifying rolling bearing damage locations based on optimal informative frequency bands are limited by the accuracy of frequency band segmentation strategies and the characterization capability of frequency band quantitative indices. To address this issue, a rolling bearing fault diagnosis method based on a multi-index fused informative frequency band spectrum was proposed. First, an adaptive frequency band division strategy was designed to segment and recombine the autoregressive power spectrum of the original vibration signal, thereby obtaining informative frequency bands with different center frequencies and bandwidths. Second, negentropy and the Gini coefficient were weighted and fused to construct a frequency band quantitative index that is sensitive to periodic impulse characteristics and has good noise robustness. Finally, the index was used to select the optimal informative frequency band from the band group to identify the bearing fault location. Simulated signals and measured data were used to validate the proposed method, and comparisons were made with the fast kurtogram, Infogram, and Autogram methods. The simulation results show that the fault characteristic frequency and its second harmonic in the optimal informative frequency band selected by the proposed method are clearly visible, whereas the comparison methods can only identify the fundamental fault characteristic frequency, with the second harmonic submerged by noise. The measured data show that the proposed method can extract the fault characteristic frequency and its harmonic components more clearly, and the diagnostic results are consistent with the actual damage location of the bearing. The proposed method can effectively improve the accuracy of rolling bearing fault diagnosis.