Controlled laboratory-scale column deposition experiments were conducted using a well-characterized mutant of the Escherichia coli (E. coli) K12 strain to obtain insight into the mechanisms that give rise to the observed deviation from classical colloid filtration theory (CFT). Both the suspended effluent bacteria concentration and the spatial distribution of retained bacteria were systematically measured over a wide range of solution conditions using columns packed with spherical glass beads. Calculations of Derjaguin--Landau--Verwey--Overbeek (DLVO) interaction energies based on measured cell zeta potentials indicated that the bacteria should experience considerable repulsive interaction forces when approaching the glass bead surface. In spite of these predictions, bacterial adhesion was observed even at the lowest solution ionic strength investigated (3 mM) and increased with solution salt concentration. Comparison of these results with measurements obtained using model colloidal particles (polystyrene latex microspheres) and a different microbe (Cryptosporidium parvum) suggested that another non-DLVO-type interaction may be contributing to the observed deposition behavior. Furthermore, predictions based on a discrete dual deposition mode (DDM) model disagreed with measured fractions of released cells. Taken together, the experimental and modeling results suggest that the deposition behavior of bacteria in saturated porous media is influenced by additional interaction mechanism(s) or factors not considered in classical DLVO theory, such as local charge heterogeneities of the cell membrane and surface biomolecule-specific interactions.