This paper focuses on a robustness controller and the performance of a permanent magnet synchronous machine (PMSM) with parameter uncertainty. The proposed controller is composed by a model reference adaptive current controller and a torque compensator. An internal model of the PMSM is introduced to define the transient reference behavior of the current and torque. A simple adaptation rule of the model uncertainty based on the Lyapunov theory and Symmetrical Optimum is given. With this approach the performance of the current controller and uncertainty estimation are guaranteed. For the torque estimation, the minimum switching time and the harmonics of the converter as well as the zero order hold in measurement result in steady state estimation error. In this paper, a torque estimator is given to compensate the mentioned torque error. In the end, the simulation results demonstrate the performance of the adaptive controller and the torque estimator.