Aiming at the position sensorless vector control system of surfac permanent magnet synchronous motors, an adaptive second-order sliding mode observer with online identification of motor parameters based on a super-twisting algorithm is presented. In the two-phase stationary coordinate system, the paper combines a model reference adaptive system with a second-order sliding mode based on the super-twisting algorithm. Thus the exact estimate of the back electromotive force (EMF) is achieved. The Lyapunov theory is used to prove the stability of the observer, and the adaptive law of stator resistance and rotor speed is deduced from the Lyapunov stability equation. In the synchronous rotating coordinate system, the second-order sliding mode observer is used to estimate the permanent magnet flux. Subsequently, the identified parameter is brought in the position tracking observer to estimate the rotor position. The algorithm can effectively suppress the sliding mode chattering without the use of low-pass filter and phase compensation. The rotor position measurement is not affected by the changes of stator resistance and permanent magnet flux, which owns strong robustness. The simulation results show the effectiveness of the proposed algorithm.