Finite　control　set　model　predictive　control　(FCS-MPC)　of　three-level　neutral　point　clamped　inverter-fed　induction　motor　has　received　wide　concern　recently,　thanks　to　its　low　switching　frequency　and　fast　dynamics.　However,　the　performance　largely　depends　on　the　accurate　measurement　and　parameters　which　determine　the　feedback　and　control　loops,　respectively.　Besides,　FCS-MPC　demands　a　large　number　of　resources　to　perform　the　exhaustive　search,　which　indicates　that　the　computation　burden　increases　exponentially　with　the　increasing　switching　states.　Aiming　at　improving　the　robustness　under　the　condition　of　unavoidable　measuring　noises　and　parameter　variation　as　well　as　reducing　the　computational　burden,　a　passivity-based　model　predictive　control　(PB-MPC)　scheme　is　presented　in　this　article.　For　an　ideal　system,　the　PB-MPC　has　a　similar　character　as　that　of　the　FCS-MPC.　In　real　applications,　where　noises　and　disturbances　impact　the　system　more　or　less,　PB-MPC　outperforms　FCS-MPC　due　to　the　power　shaping　and　damping　injection　inherited　from　passivity-based　control　which　ensures　the　asymptotic　stability.　A　Lyapunov　function　is　designed　to　prove　the　stability　of　the　proposed　scheme.　The　stability　and　efficiency　are　verified　by　both　simulation　and　experimental　results.