A new method based on finite element analysis and wavelet analysis for the fault diagnosis of wind turbine main bearings is presented in the paper. Firstly, take example for low speed bearing, the three-dimension models of the good and fault bearings are set up, which was discussed with using finite element method. The stress, strain and contact stress distribution were computed by this method. Then the wavelet transform of data signals collected in experiment was carried out with db6 wavelet and 4 scales on the basis of the stress wave analysis. The frequency D3 and D4 signals were reconstructed and characteristic frequencies of stress waves with the simulative faults were extracted. It has been shown that the method applying finite element analysis and wavelet analysis is an efficient method for the fault diagnosis of wind turbine main bearings.