Process-based crop models can integrate the complex interactions of soil properties, climatic conditions, management practices, and crop genetic characteristics. They differ in the way they simulate soil-plant-atmosphere processes, in detail of describing crop growth, water use and soil water balances, also in the number of parameters and inputs required (Rötter et al., 2012). The prediction uncertainty of crop models not only differs from one model to the other but also through different environmental conditions (Ramirez-Villegas et al., 2017). In fact, crop models response differently and nonlinearly to different environmental conditions. In addition to the model structure, the number of environments using for the model calibration also impact on the model response to the different environmental condition (Kassie et al., 2016). This paper aims to compare the response of DSSAT v.4.7 wheat models (CERES-Wheat, CROPSIM-Wheat and N-Wheat) with different climates using different cultivars.