Frequency of extreme weathers is projected to increase under the scenario of climate change, which can cause extensive damages on crop production. Wheat is one of the most important staple crops in China, and drought is the main climate disaster affecting its yield, particularly in northern parts of the country. The temporal and spatial changes of drought impacts under climate change on wheat are therefore a great concern. Most previous studies investigated the drought impacts through the use of climate indices, such as SPI. The climate indices usually neglect the diverse responses of different crop growth stages to a certain degree of drought shock, and they also disregard the contrasting characteristics in terms of the drought resistance between crop genotypes. This study assessed the potential yield impacts of past drought anomaly (1962-2010) on winter wheat, by using a process-based crop model CERES-Wheat. The model was driven by daily weather data and soil data at a grid scale of 50 km *50 km. The output of rainfed and fully irrigated yields were retrieved for further analysis. We used the difference between simulated irrigated and rainfed wheat yields to define a potential yield loss caused by drought. The anomaly percentage of the potential yield loss was employed to reflect the interannual and interdecadal variation of estimated impact of drought on winter wheat. We investigated the moving of drought susceptible regions by calculating a geographically drought-affecting center for each 10-year period from 1962 to 2010 (1960s: 1962-1970, 1970s: 1971-1980, 1980s: 1981-1990, 1990s: 1991-2000, 2000s: 2001-2010). Finally, the relationships between estimated yield impacts and recorded indices of atmosphere circulation was examined to understand the underlying mechanisms of past drought risks. The indices were averaged over the prior to winter wheat growing-season (June-Sept.) and the whole growing-season (Oct.-May), respectively. Our results demonstrated that simulated winter wheat yield exhibited a fast decreasing rate under the rainfied environment than that under the full irrigated condition, resulting in a slight increase of estimated potential yield loss from 1962 to 2010. These suggest that overall past climate change has caused increased drought risks for winter wheat production in China. Northwest China, especially the Loess Plain and Hexi Corridor showed the fastest increase of the drought risks among all winter wheat growing regions of China. However, the potential yield loss experienced a decrease in a few simulated time periods, such as the1960s and the 1980s, suggesting obvious temporal fluctuation and variability of the drought risks under climate change. Furthermore, the inferred primary drought susceptible areas moved toward northwest during the past 50 years, indicated by the drift of the geographically drought-affecting centers. This moving is interpreted by the increased yield loss due to drought in northwest China, especially in the Loess Plain and Hexi Corridor. The moving of the drought susceptible areas was fast in the periods of 1960s-1970s and 1980s-1990s, probably caused by the associated changes in precipitation and temperature patterns. In addition, the estimated yield loss and the moving of the drought-affecting center demonstrated significant (P<0.05) correlations to a number of atmosphere indices (i.e. North Pole Vortex, subtropical high system). A stronger subtropical high system during the winter wheat growing season associated with a west moving of the drought-affecting center. A stronger North Pole Vortex before or during the winter wheat growing season associated with a south moving of the drought-effecting center.