According to the fourth report from the IPCC was confirmed climate change and its impacts on drought, floods, health problems and food shortages. Therefore, understanding of how climate change could be important in the management of resources, especially water resources management. Atmosphere-Ocean Global Circulation Models (AOGCM) are tools for predicting the future climate variables and it must be downscaled its output for the studies on the local scale. In statistical downscaling methods, output of GCM grid was transferred to station. The accuracy of downscaling is dependent on location of weather stations in GCM grid. The main objective of this study was to predict temperature and precipitation by using the HadCM3model under the A2 emission scenario and statistical downscaling model (SDSM) to year 2099. Furthermore, relation between accuracy of SDSM downscaling model in different station of KAN which located in one grid and location and climatic conditions of each station was evaluated. The results showed at the station that mean of temperature and rain was closer than to mean of temperature and rain of HadCM3 grid, simulation were obtained with higher accuracy. Finally, temperature and precipitation for this three period (2011-2040), (2041-2070) and (2071-2099) were predicted and compared with base period (1961-2001). The results showed temperature will increase and precipitation will decrease by 2099 in KAN watershed.