This study investigated possible change points of extreme precipitation during the past 1973-2012 and the future 2020-2059 in South Korea. The Bayesian change point detection was performed based on a generalized Pareto distribution. Changes in the shape/tail index were detected to account for the changes in the extreme data. Results indicated 20 and 24 stations exhibited a change point for the observed and projected period, respectively. An abrupt increase in the 100-year design level appeared in the majority of northern part during the past period and the increase was likely to move to the southern area for the projected period.