The deposition efficiencies and reversibilities of bacterial adhesion (Staphylococcus epidermidis HBH2 169, Acinetobacter calcoaceticus RAG-1, and Streptococcus thermophilus B) on different substratum surfaces under flow were determined using real-time in situ image analysis. For each combination of strain and substratum, the Lifshitz-Van der Waals, electrostatic, and acid-base Gibbs energies of interaction were calculated from measured zeta potentials and contact angles on the basis of a DLVO approach sensu stricto and an extended DLVO approach. In order to compare the adhesion of microorganisms with that of inert spherical particles, monodisperse polystyrene latex particles were also included in this study. The depth of the secondary interaction minimum in the distance dependence of the total Gibbs energy of interaction fully governed the deposition efficiency and the desorption of both inert particles as well as of the microorganisms. Deposition efficiencies increased with increasing depth of the secondary interaction minimum, while desorption decreased with increasing depth. Initial desorption was always higher than final desorption, but nevertheless particle and bacterial adhesion remained reversible. The reversibility observed partly represented the real thermodynamic reversibility of the process. However, collisions between flowing and adhering particles and bacteria also contributed to the reversibility observed. The DLVO approach sensu stricto did not adequately explain the phenomena observed on hydrophobic fluoroethylenepropylene substrata, because in the sensu stricto approach long-range hydrophobic attraction due to acid-base interactions is not included. For hydrophobic interacting surfaces, including bacterial cell surfaces, long-range hydrophobic attraction has to be accounted for, yielding relatively high deposition efficiencies and low desorption rates. Summarizing, it can be concluded that initial bacterial adhesion can be explained in terms of overall physicochemical surface properties and that it is mediated by reversible, secondary minimum DLVO interactions.