This paper deals with a new systematic multi-model controller design for nonlinear systems. The design of local controllers based on performance requirements is incorporated with the concept of local models selection as an optimization problem. Gap metric and stability margin are used as measuring tool and operation space dividing criterion, respectively. The developed method provides support to design a simple structured multi-PI controller which guarantees both robust stability and time-domain performance specifications. The main advantages of the proposed method are avoiding model redundancy, not needing a priori knowledge about system, having simple structure, and easing the implementation. To evaluate the presented multi-model controller design procedure, three benchmark nonlinear systems are studied. Both simulations and experimental results prove the effectiveness of the proposed method in set-point tracking and disturbance rejection.