It has been widely recognized, both in classical and in modern studies, that the Lisbon earthquake of 1755 was a multiple event, composed of three shocks separated by a few minutes (see, e.g., Reid, 1914). Attempts to constrain the location of the source have led to a diversity of proposals, reflecting apparent contradictions in the data. The tsunami and damage along the south and southwest Iberian coast and in Morocco favor an offshore source, whereas the presence of an additional zone of strong shaking in the Lower Tagus Valley (LTV), near Lisbon, favors a more northerly location. By combining the contemporary accounts with intensity data from other earthquakes, we favor a compound source with a large distance between the faults. We propose that, although the mainshock was offshore, the resulting stress changes induced the rupture of the LTV fault, at a distance on the order of 350 km (but subject to large uncertainty in the offshore location), a few minutes after the mainshock. We favor this model, rather than site effects causing high intensities in the Lisbon area, because the highest intensities show a negative correlation to soft soil. Several other phenomena described in the eyewitness accounts can also be explained by the local rupture now proposed, such as a tsunamilike wave in the Tagus River, ground deformation affecting the course of the Tagus River, and the spatial pattern of damaging aftershocks. Recognition of this “missing” episode of rupture on the LTV fault significantly changes the hazard estimate for the Lisbon area.