This work presents a new methodology for controlling processes that exhibit multiple steady states. The proposed approach is based on a Model Predictive Control (MPC) framework, where the dynamics of the process are modeled by a Radial Basis Function (RBF) neural network. The innovative non-symmetric fuzzy means algorithm is employed in order to train the RBF network. The proposed methodology is applied to the control of a non-isothermal Continuous Stirred Tank Reactor (CSTR) that exhibits three steady state points. The results show that the proposed controller can drive the CSTR through the entire operating region, including the unstable steady state point, around which the control task is rather challenging.