Fourty percent of all crops grown in the world today are grown using irrigation, and shifting precipitation patterns due to climate change are viewed as a major threat to food security. This thesis examines, in the framework of the MIT Integrated Global System Model, the potential impacts of climate change on crop water stress and the risk implications for policy makers due to underlying uncertainty in climate models. This thesis presents the Community Land Model - Agriculture module (CLM-AG) that models crop growth and water stress. It is a global generic crop model built in the framework of the Community Land Model and was evaluated for maize, cotton and spring wheat. A full climate model, the IGSM-CAM, was first used to force CLM-AG and show the regional disparity of the response to climate change. Some areas like the Midwest or Equatorial Africa benefit from the higher precipitations associated to climate change while others like Europe or Southern Africa see the irrigation need for crops increase. The effect of a mitigation policy appeared contrasted, as water-stress for some areas (including Europe and Africa) is increased if greenhouse gases emissions are limited while for other areas (Central Asia, United States) it is reduced. A second analysis was carried in Central Zambia using uncertainty ensembles. The ensembles demonstrate the notable extent of the uncertainty stemming from different climate sensitivities and different regional patterns in climate models. Crops are impacted differently but a consistent result is that climate mitigation policies reduce uncertainty in crop water stress, making it easier to plan for any anticipated future climate change.