Radioligand binding studies have shown that AMPA receptors exist in two variants that differ about twenty-fold in their binding affinities, with brain receptors being mainly of the low-affinity type and recombinantly expressed receptors having almost exclusively high affinity. However, the physiological correlate of high- and low-affinity binding is not yet known. In this study we examined if physiological experiments similarly reveal evidence for two distinct receptor variants. We therefore measured equilibrium desensitization by glutamate and determined IC(50) values for neuronal receptors and for the homomeric receptors GluR1-4 expressed in HEK293 cells. Contrary to the prediction that these IC(50) values exhibit large differences commensurate with those of high- and low-affinity binding, values for homomeric receptors (1-18 microM) were on an average not different from those of neuronal receptors (3-10 microM). Moreover, simulations with kinetic receptor models suggest that the IC(50) values for neuronal and recombinant receptors correspond to the binding affinity of the low-affinity receptor variant. These findings indicate that the high-affinity binding measured in heterologous expression systems represents an immature receptor variant that does not contribute to the currents recorded from these cells, and that the functional low-affinity receptors are present in such small number that they are effectively masked in binding assays by the high-affinity receptors. Thus, in order to compare experimentally determined saturation binding profiles with those predicted by kinetic receptor models and with dose-response curves from physiological studies, it will be imperative to develop methods for isolating first the low-affinity receptors.