Abstarct: The realization of sustainable agriculture in each region requires the observance of management issues. That is, in order to achieve optimal performance, it is essential to know all the factors affecting crop production. Estimates of the performance of strategic products such as wheat, due to the importance of economic planning it is very important. The main objective of this study was to determine the effective parameters in estimating the yield of irrigated wheat an artificial neural network model was used to predict the yield of irrigated wheat with the help of climatic parameters in Shahrood and between 1996 and 2014. In order to apply the neural network, was used Qnet 2000 software, which is able to determine the percentage of input parameters input on the output. Model inputs include minimum absolute temperature, mean of minimum temperature, maximum absolute temperature, average maximum temperature and average temperature, absolute absolute humidity, average relative humidity, absolute maximum relative humidity, average relative humidity and average relative humidity, rainfall, Maximum wind speed, maximum sunshine, number of freezing days, irrigation water depth and evapotranspiration. The results of this study showed that the minimum absolute parameters of water temperature and depth of irrigation water were the most and the wind speed parameter had the least effect on the estimation of yield in the study area. Also, baxsed on the results, the neural network with sigmoid transfer function and structure 7-15-15-1 (7 input neurons, 2 intermediate laxyers and 1 output) yielded the best results so that was obtained root mean square error of RMSE = 167 kg / ha. Although the number of years in which wheat yields are less than their actual value is lower, however, the largest estimate error is related to the 8th and 10th years(increase 19.88% and 17% respectively, Relative to the real mode) , Which are part of the over the years. The results of this study showed that by removing the minimum absolute of temperature and depth of irrigation water components , root mean square error of 167 kg / ha would increase to 548 and 352 kg / ha respectively. The results of this research showed that artificial neural network could be a suitable alternative for plant simulator models in case of proper training and the use of reliable data.