The origin of the Comoros archipelago in the Mozambique channel is the subject of a longstanding controversy. This volcanic chain has been successively interpreted as a hotspot track, as built on an intraplate fracture zone or a passive margin, or as the northern limit of a diffuse deformation zone in the Nubia-Somalia plate system, none of these interpretations being entirely satisfactory. It is also possible that the volcanism of the Comoros is a branch of the East African Rift System, and delineates the still poorly constrained northern boundary between the Lwandle and Somalia plates. To better understand the origin of the Comoros archipelago, we combined a formal stress inversion of earthquake focal mechanisms and deformation structures (faults and dykes) observed on three islands (Mayotte, Anjouan, and Mohéli) with a morphologic study of the repartition of onshore and offshore volcanic vents in the area. Earthquake focal mechanisms and field-observed deformation structures both yield a generalized strike-slip regime with a maximum horizontal stress oriented NW-SE. In tandem, the study of topographic and bathymetric features shows that volcanic vents follow a preferential repartition along two trends: N105 ± 10° lineaments up to 100 km-long, and N145 ± 10°, less than 60 km-long lineaments organized in an en échelon fashion. Earthquakes, deformation structures, and volcanic vents all consistently designate the Comoros archipelago as an E-W elongated, 150 km-wide zone of right-lateral tear in the lithosphere. Because the clockwise rotation of Somalia relative to Lwandle requires the existence of such a right-lateral shear zone, we interpret the Comoros archipelago as the northern boundary between the two plates. This plate boundary is still in an immature state, which is why it remained undetected until now.