The paper introduces a new approach to detect the fault of bearing based on wavelet grey moment vector and hidden Markov modeling (HMM). Because of non-stationary characteristics of vibration signals of faulty bearings, we propose a new method to extract the wavelet grey moment vectors from these signals. The grey moment vectors are used as feature parameters to train HMMs to establish the database. Fault modes of bearings can be identified by select the HMM with the highest probability. The experimental results show that the proposed approach is effective and accurate to detect the faulty bearing for every single fault.