In order to evaluate annual maximum wind speeds due to typhoons, a new probabilistic procedure was proposed in which differences in the average time between observations and calculations were considered. Comparing full-scale observations and wind speeds calculated from a typhoon model, it was found that the average time of the calculated wind speeds was much more than 10 minutes, and the calculated wind speeds did not include the small scale variation which was found in the observations. By introducing the probabilistic procedure into the Monte-Carlo technique, the calculated probability distribution for 10 minutes of mean annual maximum wind speed had better agreement with observations than those by the conventional approach which underestimated annual maximum wind speeds.
