Two integrated multilinear model predictive control (MLMPC) algorithms are proposed for nonlinear chemical processes. The gap metric and the gap metric stability margin are employed to select local linear models and design local MPC controllers. Thus, the local stability and desired closed-loop performance can be incorporated into the model bank selection process. After that, a gap-metric-based weighting method is used to combine the local MPC controllers into a global MLMPC controller for the nonlinear process. Therefore, the local model selection, the local controller design, and the local controller combination are all completed according to the gap-metric-based criteria. Close connections are established among the three key elements of the multilinear model predictive control approach. Thereby the design of a MLMPC controller is more systematic, which is found to improve the accuracy and robust performance of a MLMPC controller. Since the gap metric does not consider constraints and the use of linear models in the multimodel approach may not lead to a globally stable control system, an additional simulation-based criterion is employed to evaluate the overall closed-loop performance. A single-input, single-output (SISO) and a multi-input, multi-output continuous stirred tank reactor (MIMO CSTR) processes are studied to demonstrate the effectiveness of the proposed algorithms.