Combining the wavelet transform and approximate entropy theory,according to the characteristic that the approximate entropy will become larger with the increase of signal complexity,it is applied to the analysis of acoustic emission signals of the bearing. In the test,through collecting acoustic emission signals under such working conditions as different stimulated defects,speeds or loads,then,the wavelet packet de- noise and transform is applied to this signals,and the maximum energy subband is found,and the approximate entropy calculation is carried out. Based on the analysis of the signals collected,it is concluded that the approximate entropy will become larger with the increase of radial load and speed in fault bearings while this phenomenon will not appear in the normal bearing. Thus a good diagnosis effect is achieved.