To discriminate between two general models of antagonism (participation and allosteric), an opioid antagonist lacking the basic nitrogen of tyramine was designed and characterized. Cyclo-[Tyr(Me)2-Tic-], the diketopiperazine of 2,6-dimethyltyrosyl-1,2,3,4-tetrahydroisoquinoline-3-carboxylic acid, is a partially rigid opioid antagonist; its pA2 (5.8) is one smaller than that of N,N-bisallyl-enkephalin but it has a very high binding affinity (10 nM) and has a delta selectivity (66 with respect to the binding to mu receptors) higher than that of naltrindole. The conformational state of this diketopiperazine, studied under a variety of solvent and temperature conditions by NMR and molecular dynamics, can be described in terms of only three conformers whose relative populations vary widely with solvent. Only one of the three conformers, characterized by a 90 degree arrangement of the aromatic rings of Tyr(Me)2 and Tic similar to those of rigid agonists and of the bioactive conformation of the corresponding linear antagonist, is consistent with the antagonist activity. This finding favors the participation model among the general mechanisms proposed to explain antagonism. Due to the simple composition of the conformational mixture and to the rigidity of the molecule, it is possible to propose a quantitative explanation for the discrepancy between the very high binding affinity (10 nM) and the fairly small in mouse vas deferens value (1.5 microM).