Wheat is an staple food in many countries and is grown in varied climates from humid to semiarid areas. Crop-Weather models have been used to predict wheat yield in many countries. In this paper, an attempt was made to predict wheat yield at Gharakhil (North of Iran). Thirteen years (1982/83- 2000/2001) data were used for model making, and five years (2002-2006) data were used for testing the model. Full model and regression analysis were performed, using SPSS computer package. The full model regression, with minimum parameters (Y=11749.741+161.884 X 1+8.973 X2-41.407 X3+23.707 X4-15.595 X 5 - 168.389 X6) can be used to predict the wheat yield products (Y) one month before harvesting at Gharakhil. The six predictors are: the average of maximum temperature in January (X 1), the total sunshine hours in February (X2), the total rainfall in February (X3), the average of minimum relative humidity in March (X4), the total rainfall in May (X5 ), the average of minimum relative humidity in May (X6). Although the determination coefficient of the model is high ( R2=0.91), there are some differences between the observed and predicted yield product from one year to another in the data set. The RMSE was calculated as 408 kg.ha-1 , which shows that non weather factors might influence the yield product. The results show that sensitive stages of wheat growing are affected by meteorological factors. These stages are flowering, milk and wax maturity. Germination and tillering stages have been affected the least by weather parameters.

Keywords