The basic frequency identification for existing masonry load-bearing walls is important for structural safety assessments.To mitigate the noise impacts on the dynamic test signals and the frequency identification results,artificial noised signals were adopted in the analysis,so as to obtain optimized parameters for the wavelet threshold de-noising method.Then pseudo-static tests were conducted on a wall specimen and the acceleration signals under ambient excitations were recorded.Moreover,the power spectrum density (PSD) functions of the original signals and wavelet threshold de-noised signals were calculated,and then the basic frequencies were identified respectively by using the peak-picking method.It is shown that the variability of the identified frequency decreases by the wavelet threshold de-noising with optimized parameters.By mitigating the noise impacts,the proposed method can improve the effectiveness of the frequency identification for masonry load-bearing walls with damages,and then give fundamental data for the safety assessment of existing masonry structures.