We reexamine original felt reports from the 1811-1812 New Madrid earthquakes and determine revised isoseismal maps for the three principal mainshocks. In many cases we interpret lower values than those assigned by earlier studies. In some cases the revisions result from an interpretation of original felt reports with an appreciation for site response issues. Additionally, earlier studies had assigned modified Mercalli intensity (MMI) values of V-VII to a substantial number of reports that we conclude do not describe damage commensurate with intensities this high. We investigate several approaches to contouring the MMI values using both analytical and subjective methods. For the first mainshock on 02:15 LT December 16, 1811, our preferred contouring yields M??7.2-7.3 using the area-moment regressions of Johnston . For the 08:00 LT on January 23, 1812, and 03:45 LT on February 7, 1812, mainshocks, we obtain M??7.0 and M??7.4-7.5, respectively. Our magnitude for the February mainshock is consistent with the established geometry of the Reelfoot fault, which all evidence suggests to have been the causative structure for this event. We note that the inference of lower magnitudes for the New Madrid events implies that site response plays a significant role in controlling seismic hazard at alluvial sites in the central and eastern United States. We also note that our results suggest that thrusting may have been the dominant mechanism of faulting associated with the 1811-1812 sequence. Copyright 2000 by the American Geophysical Union.