[3H]Neuropeptide Y labelled neuropeptide Y receptors in rat forebrain membranes as a homogenous class of high-affinity sites. Between 80 and 85% of these receptors showed high affinity for Y1-selective antagonists such as (R)-N2-(diphenylacetyl)-N-[(4-hydroxyphenyl)methyl]-D-arginine amide (BIBP3226). While competitive in functional studies, BIBP3226 produced parallel shifts of the Scatchard plots of [3H]neuropeptide Y saturation binding in rat forebrain membranes. Mechanisms which are routinely invoked to explain non-competitive binding do not apply to BIBP3226. Wash-out experiments, involving successive treatment of the membranes with BIBP3226, buffer (wash-out step) and [3H]neuropeptide Y, argue against irreversible or a pseudo-irreversible binding of the antagonist. Allosteric inhibition is also unlikely since BIBP3226 did not affect the rate of dissociation of [3H]neuropeptide Y in isotope dilution experiments. The non-hydrolyzable guanine nucleotide, 5'-guanylylimidodiphosphate (Gpp(NH)p), abolished the binding of [3H]neuropeptide Y and increased its rate of dissociation in isotope dilution experiments. This suggests that the initial [3H]neuropeptide Y-receptor association is a low affinity process and that the observed binding of [3H]neuropeptide Y is related to the formation of a ternary [3H]neuropeptide Y-receptor-G protein complex. Two- or even multistate models (in which BIBP3226 could potentially behave as an inverse agonist) could therefore be needed to explain the non-competitive antagonism of BIBP3226 in broken cell preparations.