Frequency-intensity data from the Southeastern U.S., Central Mississippi Valley, and Southern New England are compared. They are all quite parallel to one another and consistent with a slope of about 0.57. There is no evidence for the existence of upper bounds to maximum epicentral intensity in these data sets. Linear extrapolation of the frequency-intensity data to intensities of X leads to expected probabilities for the occurrence of large earthquakes. The largest events which have occurred in these three regions are consistent with these probabilities.