AbstractThis paper introduces four downscaling techniques and one spatial downscaling technique (kriging) used in the ADAPT project. Statistical methods 1 and 2 are simple methods that correct future projections based on annual and monthly averages, respectively. Method 3 adds to method 2 a correction across all spatial raster cells. Method 4 is a method that corrects the global circulation model (GCM) outputs using the standard deviation of the measure data. All downscaling techniques were applied to precipitation and temperature projections for the baseline period 1961-90 in seven river basins: Mekong (shared by China (Yunnan Province), Myanmar, Laos, Thailand, Cambodia and Vietnam), Rhine (shared by Germany, the Netherlands and France), Sacramento (California, USA), Syr Darya (shared by Kyrgyzstan, Uzbekistan, Tajikistan and Kazakhstan), Volta (Ghana), Walawe (Sri Lanka), and Zayandeh (Iran). The results were compared in graphs and tables using statistical test parameters. Based on the root mean square errors, methods 1 and 2 performed worst and methods 3 and 4 performed best. All four methods were able to correct the mean values of the GCM projections for temperature and precipitation to the measure data. However, after correction, the corrected extreme values from methods 1 and 2 in particular do not match the measured values as well. It is therefore recommended to use either method 3 or method 4 for downscaling.