The persistence of chronic pain and continuing overdose deaths from pain-relieving opioids targeting μ opioid receptor (μOR) have fueled the need for reliable long-term analgesics which use different targets and mechanisms. The δ opioid receptor (δOR) is a potential alternative target for non-addictive analgesics to alleviate chronic pain, made more attractive by its lack of respiratory depression associated with μOR agonists. However, early δOR full agonists were found to induce seizures, precluding clinical use. Partial δOR agonists may offer more controlled activation of the receptor compared to full agonists, but the development of such ligands has been hindered by uncertainty over the molecular mechanism mediating partial agonism. Using a structure-based approach, we explored the engagement of the sodium binding pocket in δOR and developed a bitopic ligand, C6-Quino, predicted to be a selective δOR partial agonist. Functional studies of C6-Quino revealed that it displayed δOR partial agonist activity at both G-protein and arrestin pathways. Its interaction with the sodium pocket was confirmed and analyzed using a single particle cryo-EM. Additionally, C6-Quino demonstrated favorable chemical and physiological properties like oral activity, and analgesic activity in multiple chronic pain models. Notably, μOR-related hyperlocomotion and respiratory depression, and δOR-related convulsions, were not observed at analgesic doses of C6-Quino. This fundamentally new approach to designing δOR ligands provides a blueprint for the development of partial agonists as safe analgesics and acts as a generic method to optimize signaling profiles of other Class A GPCRs.