According to the traditional receptor model, competitive antagonists share with agonists the ability to bind to a common site on receptors, but they are different from agonist in that they cannot trigger the biological response-i.e., they lack intrinsic efficacy. Recent findings extend the model by indicating that not all antagonists display an intrinsic efficacy of zero but that some display inverse agonism . In the present study we studied the inverse agonism at A₁ adenosine receptors in membranes prepared from rat cerebral cortex. Eight commercially available A₁ adenosine receptor antagonists (CGS-15943, ADPX, CPT, DPCPX, DPX, N-0840, PACPX and 8-PT) were screened for inverse agonism by measuring the extent of [³⁵S]guanosine-5 -(γ-thio) triphosphate ([³⁵S]GTP_γS) binding to G proteins. The agonist-induced stimulation of [³⁵S]GTP_γS bindings was completely blocked in the presence of A₁ adenosine receptor antagonists. Under optimal conditions, two types of antagonists could be distinguished. Seven antagonists including DPCPX decreased the basal [³⁵S]GTP_γS binding in the absence of agonist, displaying inverse agonist activity. One (CGS-15943) had no effect on the basal bindings. N-ethylmaleimide treatment reduced the basal bindings as well as agonist-mediated stimulation of [³⁵S]GTP_γS bindings, indicating that a substantial amount of this binding reflects an activated state of the C proteins. In good agreement with these findings, 0.1 mM GTP decreased the apparent affinity of the receptors for the agonist PIA, increased that for DPCPX, and had no effect on that for CGS-15943.