The steps of method of rolling bearings fault diagnosis were summarized based on wavelet packet energy feature and BP neural network, and the principle of wavelet packet and the connection between vibration of fault rolling bearings and signal were illuminated. On this basis, wavelet packet energy feature was extracted to construct characteristic vector, and a rolling bearings fault diagnosis experiment was designed to verify this method. The conclusion indicated that BP neural network has possessed good capability of identification with reasonable design and proper training. The results showed that it is feasible to implement fault vibrating diagnosis of rolling bearings with wavelet packet energy feature and BP neural network.