In this paper, a model predictive control (MPC) strategy to drive an induction machine (1M) using a three-level neutral point-clamped (3L-NPC) converter is investigated. The proposed strategy is a space vector modulation based MPC which controls the torque and flux of the 1M by using a cost function focusing on the average tracking error of the stator flux linkage vector. Simulation results show an improved steady-state performance maintaining the fast dynamic response of the standard finite control set MPC. The control strategy works appropriately with a relatively low sampling frequency of 1 kHz, and it exhibits zero average tracking error in the controlled variables.