The main goal of this work is to investigate the application of constrained model predictive control (CMPC) for a pilot hydrotreating plant. The pilot plant used in this study is a catalyst testing unit operated for desulphurization of atmospheric residue feedstock and using, the same catalyst and operating conditions of actual refinery hydrotreating processes. The operational goal in such reactors is to maintain the sulphur content of the product at a desired limit. Degree of desulphurization is highly sensitive to reaction temperature and is affected by a number of parameters such as H-2/oil ratio and catalyst deactivation. A CMPC controller has been experimentally implemented to optimize the reaction zone temperatures that would result in achieving the desired degree of desulphurization. Selected manipulated variables are system pressure and three reactor skin temperatures. A significant improvement in controller performance is obtained with the CMPC strategy, when compared with the existing PID-based control structure. This study highlighted the main factors affecting the operation and dynamics of such processes. The pilot plant can be considered as a test bed for implementing online optimal control and nonlinear control strategies for the catalytic hydrotreating units in refineries.