We report here the characterization of the arginine binding site(s) and corroborative neurophysiological studies. Binding of L-[3H]arginine to Fraction P2 from taste epithelium was measured by a modification of the method of Krueger and Cagan. Parameters for measuring maximal binding activity were established for both duration of incubation and pH of medium. At pH 7.8, the apparent single rate constant for association (kobs) at 4 degrees C was 4.72 x 10(+5).M-1.min-1. Dissociation was more complex, yielding two rate constants of 1.77.min-1 and 8.34 x 10(-3).min-1. These data suggest the presence of two affinity states for L-arginine. The KD values as calculated from the ratio k-1/k+1 were 1.3 x 10(-6) M and 1.8 x 10(-8) M. Homologous inhibition studies of L-arginine binding were not fit by a simple mass action relationship (Hill Coefficient 0.79), but were best fit by a two-site model with IC50 values of 1.6 x 10(-6) M for the high affinity state and 9 x 10(-4) M for the low affinity state. Multiunit neural recordings examined the stimulatory effectiveness of a number of guanidinium-containing compounds. Compared with L-arginine, only L-arginine methyl ester and L-alpha-amino-beta-guanidino propionic acid (L-AGPA) were effective stimuli. Cross-adaptation experiments demonstrated that at 10(-4) M L-arginine methyl ester, L-AGPA and, to a lesser extent, D-arginine were effective cross-adapting stimuli to 10(-6) M L-arginine. In competition binding studies L-arginine methyl ester, L-AGPA and D-arginine also inhibited binding of L-[3H]arginine (10(-6) M), but each recognized only one affinity state. Inhibition by the poorly cross-adapting stimuli L-glutamate, glycine and L-alanine occurred only above 10(-3) M, indicating that the binding sites for L-arginine are selective. These studies suggest that there are at least two affinity states of L-arginine binding, that the binding sites are specific, and that effective agonists of L-arginine receptors must contain a guanidinium group and an unblocked L-alpha-amino group.