Three different gas-liquid-solid reactor configurations have been used to investigate the performance of a P-doped NiMo/Al2O3 catalyst in the hydrodesulfurization of dibenzothiophene. The commonly used millipacked bed reactor with 250-500 mu m catalyst particles diluted with 125 mu m inert particles, a micropacked bed reactor with 55-90 mu m catalyst particles, and a slurry reactor with 150-250 mu m catalyst particles were used in the catalyst performance testing program. It appeared that the inherently small particle size in the packed beds causes the hydrodynamics to be dramatically different compared to the industrially applied trickle-bed reactors. For particles smaller than typically 2 mm the capillary forces predominate over the viscous and gravitational forces, in sharp contrast to large-scale industrial reactors. Since the gas flow follows preferential pathways through beds consisting of small particles, the poor radial dispersion of the gaseous components can cause mass-transport limitations, even for a rather slow reaction such as the hydrodesulfurization (HDS) of dibenzothiophene, as a result of the strong inhibition by the reaction product H2S. An adapted criterion is proposed for estimation of the contribution of poor radial dispersion in catalyst performance testing.