Aiming at online identification of permanent magnet synchronous motor (PMSM) parameters, a small speed disturbance injection method based on forgetting factor recursive least square (FFRLS) is proposed for online identifying both electrical and mechanical parameters of PMSM. The estimation can be separated into two parts. Electrical parameters such as stator resistance, inductance and flux linkage can be identified in real time by using FFRLS. Considering the operation condition of a constant speed, it is hard to identify the rotational inertia online because the derivative of angular velocity is nearly zero. To solve the problem, the small speed disturbance injection method is used to identify the mechanical parameters. Simulation results show that the proposed method can identify the electrical and mechanical parameters of PMSM accurately and rapidly.