Taiwan is situated in the tropical and subtropical region with frequent typhoons and earthquakes. The characteristics of the non-uniform spatial and temporal distribution and the steep terrain make soil erosion a serious environmental issue in Taiwan. The Universal Soil Loss Equation (USLE) is a commonly used formula to estimate soil erosion. In this study, Rainfall erosivity index (R) was re-evaluated because climate change has influenced rainfall distribution and seriously affected the estimate of soil erosion. Nowadays, climate change simulation is used to evaluate long-term trend, but not for extreme event. As a result, this study discussed erosivity index (R) in the short-term, mid-term and long-term periods in the future. This study discussed and compared two climate scenarios (SRES-A2, B2) rainfall data which were obtained by various GCMs (CCSRNIES, CGCM2, EH4, GFDL and HADCM3) released by Intergovernmental Panel on Climate Change (IPCC). Rainfall erosivity index (R) of climate scenarios was re-evaluated from annual rainfall derived from GCMs by the relation between annual rainfall and R. The rainfall erosivity indices of short-term, medium-term and long term periods in the future were gained individually. Results showed that R increased in the future; the increase of R in southern Taiwan is larger than that of other place. Variation of R based on various GCMs is larger along the time period. The erodent map of Taiwan was obtained using Kriging estimation, which can be utilized for long-term watershed management and the estimation of soil loss.