It is now well established that the early continental crust was formed by melting of basaltic lithologies such as amphibolite and eclogite. However, considerable uncertainty surrounds the geologic environment in which melting took place. Commonly invoked options range between melting at the underside of oceanic plateaus above mantle plumes or melting of oceanic lithosphere during shallow subduction. Distinguishing between these scenarios has important implications for the early evolution of continents. We use the first eclogites discovered from the North Atlantic craton (NAC) to constrain the formation of the deep root to this continent. Late Archean eclogite xenoliths (2.7 ± 0.3 Ga) from a kimberlite in West Greenland are broadly coeval with a major regional episode of tonalite-trondhjemite-granodiorite (TTG) magmatism. Major and trace element systematics of the eclogites reveal a highly refractory character that is mirrored by NAC peridotites. Moreover, the refractory eclogites define a complementary relationship to the Late Archean TTG granitoids from the NAC, and their elevated garnet δ18O values along with negative Eu anomalies suggest seafloor-altered oceanic crust as the most viable eclogite protolith. These results from Greenland provide strong support for a model in which early continental crust grew by melting of basaltic slabs in subduction zones, where tectonic stacking of down-going oceanic lithosphere provided the mechanism that coupled formation of cratonic crust and mantle.