This paper describes the application of predictive control to the temperature regulation in agricultural processes. The main aim is to achieve climatic control of a greenhouse built in the Institute for Horticultural and Agricultural Engineering (ITG) of the University of Hannover (Germany). The MPC controller implemented here has the characteristic to take in account constraints in both manipulated and controlled variables using on-line linearisation, with very low computational burden. By means of a real time experiment, important advantages of the MPC algorithm are demonstrated, performance and mainly saving energy, using a soft optimal control effort. This controller has been compared with an adaptive PID controller.