In the 21st century, climate change in the dry lands is considerate to be one of the greatest environmental threats to the world. This study presents the projection of future changes in maximum temperatures under A2, A1B and B1 SRES of the CGCM3 outputs using the statistical downscaling model (SDSM) in the 7 selected station over Iran during the two future periods (2041-70 and 2071-99). For this purpose, after examining the ability of SDSM in simulation of the basic period climate (1981-2010), the daily maximum temperature for future decades was downscaled by considering the uncertainty in 7 synoptic stations as climatic representatives of Iran. In uncertainties analysis related to model-scenarios, it was found that CGCM3 model under scenario B1 had the best performance in the simulation of future maximum temperatures among all of the scenarios. The findings showed that average maximum temperature at study stations over Iran will be increased between 0.3 and 3.5 degrees C. Also this maximum temperature changes will be more severe increased based on scenario A2 compared to other scenarios of the CGCM3 model.