Adhesion G-protein-coupled receptors (aGPCRs) are characterized by the presence of auto-proteolysing extracellular regions that are involved in cell–cell and cell–extracellular matrix interactions1. Self cleavage within the aGPCR auto-proteolysis-inducing (GAIN) domain produces two protomers—N-terminal and C-terminal fragments—that remain non-covalently attached after receptors reach the cell surface1. Upon dissociation of the N-terminal fragment, the C-terminus of the GAIN domain acts as a tethered agonist (TA) peptide to activate the seven-transmembrane domain with a mechanism that has been poorly understood2,3,4,5. Here we provide cryo-electron microscopy snapshots of two distinct members of the aGPCR family, GPR56 (also known as ADGRG1) and latrophilin 3 (LPHN3 (also known as ADGRL3)). Low-resolution maps of the receptors in their N-terminal fragment-bound state indicate that the GAIN domain projects flexibly towards the extracellular space, keeping the encrypted TA peptide away from the seven-transmembrane domain. High-resolution structures of GPR56 and LPHN3 in their active, G-protein-coupled states, reveal that after dissociation of the extracellular region, the decrypted TA peptides engage the seven-transmembrane domain core with a notable conservation of interactions that also involve extracellular loop 2. TA binding stabilizes breaks in the middle of transmembrane helices 6 and 7 that facilitate aGPCR coupling and activation of heterotrimeric G proteins. Collectively, these results enable us to propose a general model for aGPCR activation.