Highly selective nonpeptide ligands with potent delta opioid receptor antagonist activity have been developed using the message-address concept. This approach envisaged the delta opioid receptor to contain two major recognition subsites; a message subsite which recognizes the pharmacophore, and an address subsite that is unique for the delta receptor type and confers selectivity. The message and address components of the delta-selective enkephalins were postulated to be Tyr1 and Phe4, respectively, with Gly2-Gly3 functioning as a spacer. The message component of the target compounds in this study was derived from naltrexone and related structures. An indole system was fused to the C ring of naltrexone as a mimic of the address component. The benzene moiety of indole was viewed as the delta address component, mimicking the phenyl group of Phe4, and the pyrrole portion was used as a rigid spacer. Members of the series (1-23) were evaluated for opioid antagonist activity on the guinea pig ileum (GPI) and mouse vas deferens (MVD) preparations. Naltrindole (NTI, 1) was the most potent member of the series, with Ke values of approximately 0.1 nM at delta receptors. The antagonism by NTI was approximately 220- and 350-fold greater at delta than at mu and kappa opioid receptors. The binding of NTI and selected members of the series to guinea pig brain membranes was qualitatively consistent with their pharmacologic antagonist activity profiles in the MVD and GPI, but the Ki values were not in the same rank order. The selectivity of NTI arises mainly as a consequence of increased affinity at delta receptors. Thus, the Ke and Ki values of NTI were 1/530 and 1/90 that of the delta antagonist enkephalin analogue, ICI 174864. In contrast to NTI, ICI174864 derives its selectivity through greatly decreased recognition at mu and kappa receptors. The implications of the high affinity and selectivity of NTI as a consequence of its conformational rigidity are discussed. It is suggested that any attempt to model a receptor-bound conformation of an opioid peptide should consider affinity and potency at multiple receptor sites rather than selectivity alone.