A method based on the DempsterShafer (DS) evidential theory was pr es ented and applied to diagnose broken rotor bars of induction motors. Wavelet pac kets decomposition of the threephase stator current was carried out according t o the frequency partition characteristics of the wavelet packets transformation. The fault characteristic vector (evidence) of the broken rotor bars was establi shed by the root mean square (RMS) value of the wavelet nodes coefficient; Minko waki distance measurement was used to construct basic reliability distribution f unction and to determine its value which was evidence as a result of the rotor f aults. By using the DS evidential fusion algorithm, the broken rotor bar fault s of the motor was accurately identified. The experimental results show that the proposed method can diagnose the broken rotor bars effectively.