Considering the strong noise background and lack of sample in fault diagnosis of rolling element bearing, a rolling element bearing fault diagnosis method was proposed based on morphological filter and grey incidence. Fault characteristic could be extracted by morphological difference filter without considering frequency spectrum of vibration signal. The method of grey incidence has good performance in small-sample classification. It could be used in recognition of rolling element bearing fault pattern. Firstly, fault signal was filtered by morphological filter. Secondly, the characteristic frequency unitary amplitude extracted was considered as characteristic vector. Finally, fault pattern was classified by grey incidence between unknown pattern and standard fault pattern. Experiment results show that the method is effective.