The main objective of present work is to use GIS, along with the model obtained, to find suitable areas for main crops of the province. To do this, at first the rain fed wheat yield data was collected from ministry of Jihad-agriculture data bank during the 1374- 1382. These data were organized as output matrix of network. The input matrix of this model contained, initial meteorological parameters. For training the neural network model, v years of data was defined as train file and y remain data were the test file for the model, after fitting the data to the model and obtaining Root Mean Square error (RMSE) of the difference matrix of the input data, we find that when we used the 6 meteorological parameters (annual rainfall, number of days with thermal and cold stress, rainy days, moisture available index, and mean daily temperature) the model produced minimum RMSE amount. So we find the best model using neural network software and climatic data to predict wheat yield.In the second stage, we obtained the agroclimatic maps of Wheat using GIS. To do this, at first we should prepared the wheat climatic requirement maps. These layers were moisture available index, rainfall, rainy days, days with thermal and cold stresses and temperature maps. Finally after preparing these layers according to climatic requirements of wheat, we used IDRISI software and overlay model to produce climatic potential maps of the province for rain fed wheat.