An envelope analysis approach based on spectral kurtosis and complex shifted Morlet wavelet in the diagnostics of rolling bearings is introduced in this paper. Both the optimal center frequency and bandwidth of the band pass filte rin envelope analysis can be obtained adaptively, which permits a simple implementation of the envelope analysis. However, the approach involves an intensive computing in filter banks design and filtering for envelope detection. To solve the problem, an improved computing scheme is presented. Through it, the total computation of band pass filter and spectral kurtosis for envelope detections can be reduced dramatically based on the relation of the filtering and results of the spectral kurtosis to filters with an overlap filtering frequency band atdifferent filter banks. The present scheme makes the approach more suitable to applications. Simulation and test verify the validity of the present scheme posit ively.