The activity of a hydrodesulphurization catalyst loaded in an industrial hydrotreater is studied at start up and end of run. Catalyst initial and final activity was determined by performing HDS experiments at industrial conditions in a laboratory mini-scale hydrotreater. The results show that the deactivation of the catalyst samples collected from three different places of the industrial reactor do not vary significantly, the maximum difference among the catalyst samples, being less than +/- 4%. The experimentally determined deactivation level of the catalyst samples is compared with the deactivation estimated for the same industrial reactor and the same load using a hybrid neural network model trained with operational data of the industrial and the results are in close agreement. Catalyst deactivation appears to be faster for hydrogen consumption reactions than for hydrodesulphurization reactions indicating a decreasing hydrogen consumption trend with time in operation for specific sulphur content in the product. (C) 2008 Elsevier Ltd. All rights reserved.