This work presents the development of a model-based controller design called Multiple Model Predictive Control (MMPC) based on a set of linear, time-varying, state space models to regulate batch processes according to multiple, pre-specified reference profiles. Sufficient conditions for stability and boundedness of the dynamic evolution of the forced nonlinear system are provided. The performance of the MMPC design is demonstrated on a model of a batch reactor that represents the production of a polymer product.