For the design of structures against wind-induced fatigue not only maximum wind speed but also the time series of wind speed and direction are generally required. Currently, no rationale exists for selecting realistic typhoons to be used for determining such time series for the use of the design. Further, it is not known on how the selection of these typhoons affect the evaluation of the accumulated fatigue damages of structural components. Therefore, this study first proposes a concept and methodology for identifying most-likely typhoon (ML typhoon) with a given hazard level specified in terms of maximum wind speed. Then, using the ML typhoon and other likely typhoons, which together constitute a set of realistic typhoons, characteristics of the time series of wind speed and wind direction with given hazard levels are investigated. Thereafter, a sensitivity analysis is conducted in order to evaluate the variation of the accumulated fatigue damages induced by those typhoons, taking its basis with a hypothetical seismically base isolated building. It is found that significant difference in the evaluated accumulated fatigue damages are observed for typhoons with a common hazard level and the same likelihood. This implies selecting the design typhoon(s) requires more care not only for the maximum wind speed but also other characteristics of typhoons.