In view of the problem of poor accuracy of traditional recursive least square (RLS) method for PMSM parameter identification, a dynamic discount RLS method has thus been proposed for PMSM parameter identification. On the basis of PMSM mathematical model, a multi-parameter identification model is established to realize a real-time multi-parameter identification; with the dynamic discount factor introduced to RLS, the real time adjustment of estimation error helps to overcome the problems of traditional RLS data saturation and poor estimation accuracy. The simulation analysis and experimental results show that under different working conditions, the error of the proposed method in identifying motor stator resistance, stator inductance and permanent magnet flux parameters can be controlled within 1%, which is more improved than the traditional parameter identification methods in identification accuracy and rapidity.