This article investigates the design of linear and nonlinear model predictive controllers (MPCs) in order to improve the control of pasteurization temperature in a milk plant. MPC schemes required the development of a prediction model for use internally within the controller. An artificial neural network (ANN) model of the plant is established and validated. A linearized model is then obtained around the operating point from the ANN model. The linearized and the ANN models are used for prediction for the linear and nonlinear predictive controllers, respectively. The MPC responses are compared with a benchmark PID controller behaviour, the parameters of which have been tuned to minimize the same criteria as used for the predictive controllers.