Abstract. It is important that control center operators be alerted to system disturbances, including where, when and what disturbance occurs, so that proper anticipatory actions can be promptly taken if necessary, avoiding oscillation spreads in the power network. In this paper, the wavelet multi-resolution analysis based method is proposed to identify power system disturbances. Energy of wavelet coefficients are used as a criterion to choose optimal wavelet function and decomposition scale, which are then used for obtaining the maximum wavelet coefficients by identifying the frequency signals from wide area measurement system (WAMS). The maximum wavelet coefficients are then selected to be the indicators for disturbance identifying. The detailed procedure and effectiveness of the proposed method is demonstrated by simulations of a 10-machine 39-bus system.