This paper proposes an adaptive threshold estimation algorithm for power quality signal denoising based on wavelet neighboring thresholding classification. For input signals， firstly the optimal neighboring window size is proposed by modulus maximum wavelet domain. In wavelet domain， it shows that detail wavelet coefficients come as groups and have high local correlation. Then according to its corresponding neighboring threshold， each coefficient in a subband is classified as“large” or“small” category. Those“small” coefficients are set to zero; whereas those“large” coefficients are modeled as zero-mean Gaussian random variables with high local correlation and the estimations are obtained by using the minimum mean squared error criterion. The effectiveness of the proposed method is illustrated for the actual power quality signals denoising. Simulation results show that the proposed method has outperformed the other traditional adaptive denoising algorithms.