We report results that lend support to the hypothesis that extreme hurricane wind speeds are described predominantly by reverse Weibull distributions, which have limited upper tails. This is consistent with the result reported recently in the atmospheric sciences literature that, on physical grounds, hurricanes have finite maximum intensity. The sampling errors typical of our results are controlled by the relatively short length of the historical record (about one hundred years, i.e., a few tens of hurricane events for any one location) and are, in practice, independent of whether the size of the simulated data samples for any specified location is of the order of 1,000 or larger. According to our estimates, mean recurrence intervals of wind speeds corresponding to nominal ultimate wind loads specified in the ASCE standard 7-95 for extratropical storm regions on the one hand, and hurricane-prone regions on the other, are mutually inconsistent with respect to risk.
