Two back-electromotive-force-based position observers are compared for motion-sensorless synchronous motor drives: the reduced-order observer and the adaptive full-order observer. A stabilizing gain is proposed for the adaptive full-order observer, which guarantees the local stability of the closed-loop system, if the motor parameters are known. Equations for the steady-state position error and for the linearized estimation-error dynamics under erroneous parameters are derived, and the robustness of the two observers against parameter errors is analyzed and compared. The observers are experimentally evaluated using a 6.7-kW synchronous reluctance motor drive in low-speed operation and under parameter errors. The gain selection of the reduced-order observer is easier, but the adaptive full-order observer can be made more robust against parameter variations and noise.