The rectangular cavity structure was applied to edges of the rear view mirror, numerical computation was conducted for aerodynamic noises of the rear view mirror, and the results of optimized structure were compared with those of the original structure to verify optimized effects. Wind tunnel test was then conducted on the rear view mirror to verify correctness of the computational model. Sound pressure levels of each observation point between experimental test and numerical simulation were basically consistent, where only parts of the peak frequency points were different. The computational accuracy was very high. Two large-size vortexes were in this region behind the common rear view mirror. Vortexes above the rear part of the rear view mirror rotated anticlockwise, while vortexes under it rotated clockwise. There were two vortexes in the region behind the optimized rear view mirror, but the energy intensity of this vortex near the side window panel was very weak. Pressures on the front part of the optimized rear view mirror were also smaller than those of the original structure. In addition, it could be found that radiation noise distribution of the rear view mirror on the lateral window presented symmetry. When lateral window effects were considered, the aerodynamic noise was more than results without considering the lateral window effects. Total noises of each observation point without considering the lateral window effects were 56.7 dB, 59.2 dB, 58.6 dB, 58.9 dB, 62.3 dB and 63.1 dB, respectively. Total noises at each observation point with consideration of the lateral window effects were 65.3 dB, 66.5 dB, 68.7 dB, 69.2 dB, 70.1 dB and 70.8 dB, respectively. Therefore, when aerodynamic noises of the rear view mirror were computed, impacts of lateral window effects must be considered, otherwise the computation would be seriously deviated from actual situations. The frequency corresponding to maximum peaks of the aerodynamic noise after optimization approached the self-oscillation frequency of the rectangular cavity. The result indicates that aerodynamic noises in the rectangular cavity were fluid self-oscillation caused by the cavity structure. In addition, within the analyzed frequency band, the aerodynamic noises of the optimized rear view mirror were smaller than those of the original structure. The maximum decrease rate of total noises of the optimized rear view mirror was 15.62 %, and minimum decrease rate was 8.90 %. Optimized effects were very significant, especially in the low frequency bands.