Abstract: Aiming at non-stationary and modulation characteristics of gear fault vibration signals,a fault diagnosis method was proposed based on dual-tree complex wavelet packet transform and spectral kurtosis. First,the original vibration signal was decomposed into several different frequency band components by dual-tree complex wavelet packet transform,some components that have bigger correlation coefficient were de-noised by the threshold. Second,the best bandwidth and band center of band-pass filter were determined through fast kurtosis diagram of spectral kurtosis. Finally, the envelope demodulation spectrum of filter signal could be obtained by square envelope and Fourier transforms,then the fault information was effectively extracted. The analysis of the gear fault signals shows that the fault feature information of the gear can be effectively extracted to identify the fault,and the proposed method is effective and feasible.