Outputs of General Circulation Models (GCMs) are inadequate for the assessment of climate change on land-surface in regional scale.Statistical downscaling technique has been proposed as one of the tools to establish the relationship between the meso-scale GCM output and sub-grid-scale surface variables (such as air temperature).The Statistical Downscaling Model (SDSM) is applied to the Baiyangdian Basin in this study,with daily maximum and minimum air temperature as predictors.Selection of appropriate downscaling predictors is based on the relationship between meso-scale atmospheric variables and daily air temperature series in the Baiyangdian Basin,which are obtained from the NCEP re-analysis data set and observed data at meteorological stations,respectively.From the data set during the period of 1961-1990,data of the first 15 years (1961-1975) are used for calibration while the remaining data (1976-1990) are used for validation.Finally,daily maximum and minimum air temperature scenarios for future periods (2020s,2050s,and 2080s) are estimated using SDSM from output of the HadCM3 at 7 stations,and the temporal trends of daily maximum and minimum air temperature are analyzed.The results show that both daily maximum and minimum air temperature will increase under A2 or B2 scenario.Under A2 scenario,the magnitudes of increase for daily maximum and minimum air temperature are 0.37℃/(10 a) and 0.40℃/(10 a) in the plain area,whereas,these values are 0.61℃/(10 a) and 0.55℃/(10 a) in mountainous areas,respectively.Under the B2 scenario the magnitudes of increase are smaller in both mountainous area and plain area.The increasing tendencies for different areas are quite different.The trend in mountainous area is greater than that in plain area.The results show that both daily maximum and minimum air temperature in the Baiyangdian basin will arise in future.