Methods based on the use of optical biosensors have recently become available to provide a convenient means of determining the rate and equilibrium constants for bimolecular interactions between immobilized ligands and soluble ligate molecules. However, the association data that these methods provide are not always accurately described by the expected pseudo-first-order reaction mechanism, particularly when the ligand is immobilized on a dextran matrix. We show that a better description of the association data, especially at higher ligate concentrations, is achieved with a double exponential function, indicating that at least two rate-limiting processes are involved. Various models are considered in order to explain these observations: the presence of two (or more) distinct populations of immobilized ligand; a change, possibly conformational, in the immobilized ligand before or after ligate binding; or the hindrance of ligate binding to immobilized ligand. We suggest that steric hindrance caused by ligate binding to the dextran-coated sensor surface seems the most likely explanation for the observed biphasic association kinetics and that the faster initial phase should be used in oder to determine association constants that can be compared to those in solution.