G protein activation by different mu-selective opioid agonists was examined in rat thalamus, SK-N-SH cells, and mu-opioid receptor-transfected mMOR-CHO cells using agonist-stimulated guanosine-5'-O-(gamma-thio)-triphosphate ([35S]GTP gamma S) binding to membranes in the presence of excess GDP. [D-Ala2, N-MePhe4, Gly5-ol]Enkephalin (DAMGO) was the most efficacious agonist in rat thalamus and SK-N-SH cells, followed by (in rank order) fentanyl = morphine > > buprenorphine. In mMOR-CHO cells expressing a high density of mu receptors, no differences were observed among DAMGO, morphine or fentanyl, but these agonists were more efficacious than buprenorphine, which was more efficacious than levallorphan. In all three systems, efficacy differences were magnified by increasing GDP concentrations, indicating that the activity state of G proteins can affect agonist efficacy. Scatchard analysis of net agon stimulated [35S]GTP gamma S binding revealed two major components responsible for agonist efficacy differences. First, differences in the KD values of agonist-stimulated [35S]GTP gamma S binding between high efficacy agonists (DAMGO, fentanyl, and morphine) and classic partial agonists (buprenorphine and levallorphan) were observed in all three systems. Second, differences in the Bmax value of agonist-stimulated [35S]GTP gamma S binding were observed between DAMGO and morphine or fentanyl in rat thalamus and SK-N-SH cells and between the high efficacy agonists and buprenorphine or levallorphan in all three systems. These results suggest that mu-opioid agonist efficacy is determined by the magnitude of the receptor-mediated affinity shift in the binding of GTP (or[35S]GTP gamma S) versus GDP to the G protein and by the number of G proteins activated per occupied receptor.