Using Xenopus laevis oocytes coexpressing mammalian mu-opioid receptors (MORs), beta-adrenergic receptor kinase 2 (beta-ARK2) [also called G protein-coupled receptor kinase (GRK3)], and beta-arrestin 2 (beta-arr 2), we compared the rates of beta-ARK2 (GRK3)- and beta-arr 2-mediated homologous receptor desensitization produced by treatment with opioid agonists of different efficacies. The response to MOR activation was measured using two-electrode voltage clamp as an increase in the conductance of the coexpressed G protein-coupled inwardly rectifying potassium (heteromultimer of KIR3.1 and KIR3.4) channels. Treatment with opioids of high efficacy, either [D-Ala2,N-MePhe4,Gly-ol5]-enkephalin, fentanyl, or sufentanyl, produced a GRK3- and beta-arr 2-dependent reduction in response in <20 min, whereas treatment with the partial agonist morphine produced receptor desensitization at a significantly slower rate. Because GRK3 requires activation and membrane targeting by free G protein betagamma subunits released after agonist-mediated activation of G proteins, a low efficacy agonist such as morphine may produce weak receptor desensitization as a consequence of poor GRK3 activation. To address this hypothesis, we substituted GRK5, a GRK that does not require activation by G protein betagamma. In oocytes expressing GRK5 instead of GRK3, both [D-Ala2,N-MePhe4, Gly-ol5]enkephalin and fentanyl, but not morphine, produced desensitization of MOR-activated potassium conductance. Thus, mu-opioid agonists produced significant receptor desensitization, mediated by either GRK3 or GRK5, at a rate dependent on agonist efficacy.