In order to reduce the adverse effect of parameter variation in position sensorless speed control of permanent magnet synchronous motor (PMSM) based on stator feedforward voltage estimation (FFVE), multiparameter estimation using a model reference adaptive system is proposed. Since the FFVE scheme relies on motor parameters, the stator resistance and rotor flux linkage are estimated and continuously updated in the FFVE model in a closed-loop fashion, and the sensitivity to multiparameter changes at low speed is eliminated. To improve the dynamics and stability of the overall system and eliminate transient oscillations in speed estimation, a phase-locked loop like speed estimation method is proposed, which is obtained by passing the q-axis proportional integrator (PI) current regulator output through a first-order filter in the FFVE scheme. The proposed control method is similar to V/f control as in induction motors; therefore, starting from zero speed is possible. The experimental tests are implemented with 1-kW PMSM drive controlled by a TMS320F28335 DSP. The proposed sensorless scheme is also compared with the classical sliding mode observer (SMO). Experimental results show that the proposed sensorless scheme exhibits greater stability at lower speed than the classical SMO under parameter detuning. Experimental results and stability analysis demonstrate the feasibility and effectiveness of the proposed sensorless scheme for PMSM under various load and speed conditions.