In order to estimate parameters of permanent magnet synchronous motor (PMSM) on-line and real-time, an adaptive on-line identification method for motor parameters is proposed. Resistance, inductance and PM flux of PMSM are achieved at the same time in the presented model. By means of Popov’s hyper-stability theory, the model of parameter identification is built in the rotor reference frame. And, PMSM d, q-axis voltage, current and their errors are used to obtain the adaptive laws of parameters. Popov’s hyper-stability theory guarantees stability of the system and convergence of the estimated parameters under certain conditions. The simulation and experimental results illustrate the validity and efficiency of the proposed method.