Continental breakup and the transition to seafloor spreading is characterized by extensional faulting, thinning of the lithosphere and, at magmatic margins, voluminous intrusive and extrusive magmatism [1][2][3][4] . It is difficult to discriminate between different mechanisms of extension and magmatism at ancient continental margins because the continent-ocean transition is buried beneath thick layers of volcanic and sedimentary rocks 5,6 and the tectonic activity that characterized breakup has ceased. Instead, the timing of these mechanisms is inferred from theoretical models or from the geological record preserved at the fully developed, ancient rifted margins 1,5,7,8 . Ongoing rifting in Ethiopia offers a unique opportunity to address these problems because it exposes subaerially the transition between continental rifting towards the south and seafloor spreading further northward. Here we synthesize constraints on the spatial and temporal evolution of magmatism and extension in Ethiopia. We show that although intrusion of magma maintains crustal thickness during the early stages of the continent-ocean transition, subsidence of the margin below sea level, and eruption of voluminous basalt flows, is initiated by late-stage thinning of the heavily intruded, weakened plate just before the onset of seafloor spreading. We thus conclude that faulting, stretching and magma intrusion are each important, but at different times during breakup.The development of ancient magmatic rifted margins is often thought to be the result of an anomalously hot mantle, but the timing and rate of plate stretching are also expected to effect melt generation 7 . Ethiopia is an ideal study locale for continental breakup because it exposes subaerially several stages of tectonically active rift sector development from embryonic rifting in the south, to incipient oceanic spreading in Afar 9 ( Fig. 1). Seismic wavespeeds in the upper mantle beneath Ethiopia are among the slowest worldwide, with P-waves ∼6% slower than in normal mantle 10 . This observation has led recent studies of mantle structure to conclude that continental breakup is occurring above arguably the hottest mantle on Earth 10 . The ability to study the physical state of the mantle in Ethiopia is a distinct advantage over studies at ancient rifted margins, where the competing influences of elevated temperatures, small-scale convection, and a fertile mantle can be inferred only from the geological record. Furthermore, at fully developed margins the timing of extension by either magma intrusion or mechanical stretching, and the timing of eruption of voluminous basaltic flows that hinder deep seismic imaging at ancient magmatic margins (the so-called seaward dipping reflectors (SDRs)) cannot be established unambiguously, and it is here that we seek improvement.Rifting of Arabia from Africa above the hot Ethiopian mantle initiated on border faults of the Afar Depression ∼29-26 Myr ago 11 . Extension shifted thereafter to narrower zones of small-offset faults, fissural flows and ...