Opioid receptor binding and opioid-mediated inhibition of cAMP accumulation were studied simultaneously in intact NG108-15 cells. The dose-response curves for the biological response were suggestive of positive cooperativity and systematically occurred at lower ligand concentrations than those for the binding of [3H] [D-Ala2, D-Leu5]enkephalin (DADLE), which were instead shallow and suggestive of a site heterogeneity or of a cooperative phenomenon. Computer modeling of the binding isotherms revealed that the data are best described assuming two binding sites with different affinities for the agonist; the mean ratio between the DADLE concentrations yielding half-maximal occupancy of the high affinity site and half-maximal response was 1.5, but it was 36 when the fractional occupancy of the sum of the two sites was considered. On examining several opioids, no direct correlation was found between high affinity site and biological response; however, several agonists displayed different affinities for the two sites, while the antagonist naloxone and the partial agonist diprenorphine bound to them with identical affinities. Furthermore, naloxone exhibited a good agreement between half-maximal receptor occupancy and Ki in blocking the agonist response. Thus, the binding heterogeneity detectable in intact cells is agonist-specific, and suggests rather that the sites are states of an identical receptor population. When [3H]diprenorphine was used to label the opioid receptors, the competition curves for DADLE were consistent with the existence of an additional, very low affinity state undetectable by direct binding with labeled agonist and, again, not discriminated by naloxone. Multiple affinity states of the opioid receptor in intact cells may reflect its interaction with the effector system in the plasma membrane.