Through a coupled experimental/modelling approach, this study investigates the functioning of a fixed bed bioreactor of Biolite beads, inoculated with Pseudomonas putida, in order to biotreat phenol contaminated waters at low Reynolds number. In particular the coupling between water flow and biomass growth is studied through the evaluation of the relationship between bed permeability reduction and biomass content as well as hydrodynamic effects on biological kinetics. In term of bioclogging, our results showed a very sharp decrease of the relative bed permeability well correlated with measured biofilm growth (for relative permeability greater than 0.8) as well as a saturation of the bed permeability at low relative porosity. However, most of the models available in the literature failed to describe our observed results. Our work permitted to show a strongly heterogeneous longitudinal biofilm development along the bed and thus a strongly heterogeneous longitudinal occupation of the porosity that decreased rapidly at the bed entry. Furthermore, the observed measured porosity saturation was shown to correspond to a biomass distribution involving the existence of channelling, through a growth/detachment equilibrium, rather than a complete clogging of the bed reactor. Biofilm microstructure built at the pore scale by bacteria (e.g. via the modulation of EPS production) probably explains the observed permeability/porosity results in the early stage of bed bioclogging. However, the present experiments do not allow measuring such a microstructuration effect. Through the analysis of a simple 1D model written at the mesoscale, we showed that the closure laws accounting for the different biological kinetics (growth, detachment, etc.) are flow dependent, as they represent effective properties and complex biological processes averaged at the mesoscale. Given the low Reynolds number investigated in the present paper, this dependency appears not only controlled by mass transfer evolution but may also involve more complex biological regulation (biological response to physical chemical stresses). As a consequence, formulations that are strictly valid at the cell scale (e.g. Monod formulation) should be used with care if no averaging or upscaling methods are used. The main conclusion of the present work is that the knowledge of macroscopic laws, such as the permeability porosity relationship is not sufficient to account for the coupling between hydrodynamics and biomass growth in porous media. As bacterial biofilms are formed of populations able to adapt their metabolism to their evolving microenvironment, and in particular to hydrodynamic conditions, it appears that the predictive character of such models could be largely improved if these closure laws were not postulated a priori but deduced either from upscaling techniques, or, as in this paper, from ad hoc experiments performed at the global scale. (C) 2011 Elsevier Ltd. All rights reserved.