The　varying-coefficient　state-dependent　autoregressive　with　exogenous　inputs　models　are　very　useful　in　nonlinear　system　modeling　and　control.　Considering　that　state　feedback　control　strategy　with　a　series　of　variable　feedback　gains　provides　more　freedom　than　a　constant　feedback　gain　for　the　robust　predictive　controller　design,　this　article　proposes　a　robust　model　predictive　control　(RPC)　algorithm　with　variable　feedback　gains　for　output　tracking.　In　the　proposed　method,　by　defining　input　and　output　increment　sequences　of　the　system,　two　polytopic　state-space　models,　in　which　the　dynamic　behavior　of　the　system　is　wrapped,　are　constructed.　Then,　based　on　the　polytopic　state-space　models,　an　RPC　with　variable　feedback　gains　for　output　tracking　is　designed.　To　further　reduce　the　conservatism,　the　parameter-dependent　Lyapunov　functions　are　constructed　for　the　design　of　the　variable　feedback　gains　in　the　control　strategy.　The　proposed　RPC　can　expand　the　feasible　region　of　the　robust　controller　and　improve　the　control　performance.　The　simulation　on　a　continuous　stirred　tank　reactor　verified　the　feasibility　and　efficacy　of　the　proposed　method.