The recent literature contains conflicting claims about the characteristics of attached bacteria in subsurface porous media and how these characteristics affect permeability reduction. Some claim that the bacteria form continuous biofilms that restrict the pore size, while others claim that bacteria are attached in patchy aggregates that accumulate in pore throats. This contribution applies a recently developed tool from biofilm kinetics, the normalized surface loading, to interpret a wide range of experimental data from porous media experiments and biological filtration. The normalized surface loading is the actual substrate flux (i.e., rate of removal per unit surface area) divided by the minimum flux capable of supporting a deep biofilm. The analyses show that biofilms are continuous for normalized surface loadings greater than 0.1, but appear to become discontinuous for values less than about 0.25. For the low-load situation, distinguishing between continuous and discontinuous biofilms is not important when the modeling goal is prediction of substrate removal. However, the distinction is more critical when the modeling goal is to describe the spatial distribution of attached biomass and permeability loss. ¿ American Geophysical Union 1993