In the study of climate changes, Predict the future of climatic parameters is performed by general circulation models (GCMs) and emissions scenarios of greenhouse gas but The output of these models due to large-scale network, there are lack of appropriate spatial and temporal resolution in small-scale. For this purpose will be necessary to down scale the output of these models in e Station and the point using downscaling models, Which are divided into two categories: statistical and dynamical models, The statistical methods have application and more acceptable. Among statistical methods, SDSM LARS-WG models is the most valid tools in the currently for downscaling, Which in This study was to be analyzed the performance of these two models to simulate temperature and precipitation change in the basin of Urmia Lake in the North Western of Iran which is facing of environmental crisis and risk of drying the lake. Meteorological stations in evaluated include four stations Saghez, Tabriz, Khoy and Urmia which have been full statistics in the base period (1961 -1990). For Performance analysis models was used the indicators of MSE, RMSE, MAE and the coefficient of determination and correlation. The results showed that both models are more carefully simulated the temperature than precipitation, and simulated of monthly temperature and precipitation parameter, SDSM model is more successful and have less uncertainty and on the other hand has a time consuming and complex in simulation process. Whereas LARS-WG model have a more useful in rainfall period simulated and has simplicity and more speed in performance. In total according to the results of any of the models are not superior to each other in spite of differences in the simulation can be useful in climate change studies