Model Predictive Control (MPC), is popular in control system theory, due to their advantageous characteristics of constraints for the control signal and the possibility of measure indirectly the states. This paper presents the design of an MPC controller for a post-harvest refrigeration system in floriculture. The design is based on the simulation of different conditions. Different tests are carried out with mismatches in the state matrix in order to know whether the MPC is able to respond to such changes. A comparison is also made with an state feedback plus observer proving that the MPC has a better behavior against mismatches.