Analogues of the 2',6'-dimethyl-L-tyrosine (Dmt)-1,2,3,4-tetrahydroisoquinoline-3-carboxylic acid (Tic) pharmacophore were prepared to test the hypothesis that a "spacer" and a third aromatic center in opioid peptides are required to convert a delta-antagonist into ligands with delta-agonist or with mixed delta-antagonist/mu-agonist properties. Potent delta-agonists and bifunctional compounds with high delta- and mu-opioid receptor affinities were obtained by varying the spacer length [none, NH-CH(2), NH-CH(2)-CH(2), Gly-NH-CH(2)] and C-terminal aromatic nucleus [1H-benzimidazole-2-yl, phenyl (Ph) and benzyl groups]. C-terminal modification primarily affected mu-opioid receptor affinities, which increased maximally 1700-fold relative to the prototype delta-antagonist H-Dmt-Tic-NH(2) and differentially modified bioactivity. In the absence of a spacer (1), the analogue exhibited dual delta-agonism (pEC(50), 7.28) and delta-antagonism (pA(2), 7.90). H-Dmt-Tic-NH-CH(2)-1H-benzimidazole-2-yl (Bid) (2) became a highly potent delta-agonist (pEC(50), 9.90), slightly greater than deltorphin C (pEC(50), 9.56), with mu-agonism (pE(50), 7.57), while H-Dmt-Tic-Gly-NH-CH(2)-Bid (4) retained potent delta-antagonism (pA(2), 9.0) but with an order of magnitude less mu-agonism. Similarly, H-Dmt-Tic-Gly-NH-Ph (5) had nearly equivalent high delta-agonism (pEC(50), 8.52) and mu-agonism (pEC(50), 8.59), while H-Dmt-Tic-Gly-NH-CH(2)-Ph (6) whose spacer was longer by a single methylene group exhibited potent delta-antagonism (pA(2), 9.25) and very high mu-agonism (pEC(50), 8.57). These data confirm that the distance between the Dmt-Tic pharmacophore and a third aromatic nucleus is an important criterion in converting Dmt-Tic from a highly potent delta-antagonist into a potent delta-agonist or into ligands with mixed delta- and mu-opioid properties.