The recent article by Seth Stein, Joseph Tomasello, and Andrew Newman raised thought-provoking questions about one of the most vexing open issues in hazard assessment in the United States: the hazard posed by ostensibly infrequent, large, mid-continental earthquakes. Many of the technical issues raised by this article are addressed by A. D. Frankel in the accompanying comment. I concur with this, and will only address and/or elaborate on a few additional issues here: (1) Detailed paleoseismic investigations have shown that the New Madrid region experienced sequences of large earthquakes around 900 and 1450 A.D.in addition to the historic events in 1811–1812. With a repeat time on the order of 400–500 years, these cannot be considered infrequent events. Paleoseismic investigations also reveal evidence that the prehistoric “events” were also sequences of two to three large earthquakes with a similar overall distribution of liquefaction in the greater New Madrid region as produced by the 1811–1812 sequence [Tuttle et al., 2002]. And if, as evidence suggests, the zone produces characteristic earthquakes, one will not see a commensurate rate of moderate events, as would be the case if seismicity followed the Gutenburg-Richter distribution.