Speed and torque controls of permanent magnet synchronous motors are usually attained by the application of position and speed sensors. However, speed and position sensors require the additional mounting space, reduce the reliability in harsh environments and increase the cost of a motor. Therefore, many studies have been peformed for the elimination of speed and position sensors. This paper investigates a novel speed sensorless control of a permanent magnet synchronous motor. The proposed control strategy is based on the MRAS(Model Reference Adaptive System) using the state observer model with the current error feedback and the magnet flux model as two models for the back-emf estimation. The proposed algorithm is verified through the simulation and experiment.