Based on the analysis of spectral kurtosis using short time Fourier transform (STFT) and wavelet transform (WT), a new algorithm of spectral kurtosi s (SK) based on WignerVille distribution (WVD) was established. It was used as the detection tool and a basis for designing optimal detection filters that can efficiently extract the information of rolling bearing fault. The above three al gorithms were used to diagnose the fault of a rolling bearing. The results indic ate that the concentration extent of signal energy with the timefrequency (TF) method, selection of time windows and filters are dominant factors that affect t he algorithm. The analysis results are helpful to forming the theory system of S K algorithms based on the timefrequency analysis and the fault diagnosis.