Nowadays motor is the most widely used power supply equipment and drive. In order to ensure the reliable operation of the motor, it is necessary to analyze the fault mechanism of motor equipment and to do some research on the relevant technology about fault diagnosis. The motor diagnostic method based on signal processing is the way that relies on modern signal processing technology to extract fault features, and find the fault according to the characteristics. At present the fault diagnosis method based on the wavelet analysis has become a hot topic. But wavelet transform also has its own defects, such as its dependence on a priori information and its applicable limitation. However, Empirical Mode Decomposition (EMD) as a new signal processing technology, is completely driven by its original signal, doesn t need a priori base function and overcomes the deficiency of the wavelet transform. Therefore in this paper, EMD method is used in the fault diagnosis of the motor. In this study, EMD method shows a great advantage in the aspects of signal feature extraction, transformation, the reduction feasibility and fault diagnosis for induction motor at a setting speed.