Electric motors are found in many appli- ances, a common AC motor is the PMSM, which has a high reliability. Encoders are used for position and speed feedback, but add extra cost and reliability problems. Sensorless control options are therefore ex- plored as an alternative. This thesis inves- tigates implementing a Kalman filter in an existing sensorless drive structure to im- prove its performance. This is done using a laboratory setup, where a PMSM based drive system was coupled to a load motor, with an encoder for validation purposes. The PMSM is modelled in the different ref- erence frames, and a speed controller is de- signed and validated. A sensorless estima- tion structure, with two phase locked loops and a load observer is then designed and validated. This control structure is then modified to include a Kalman filter. These two estimation structures are tested and compared. Here it was concluded that the Kalman structure has improved the load transient response.