This study estimates the virtual water flows embodied in agricultural commodity trade among Chinese provinces on the base of input-output analysis combining the Chinese interregional input-output table and agricultural water consumption information. After the estimation, the intensity of virtual water flow (the proportions about virtual water outflow and inflow against local water resources) and scale of virtual water inflow (the proportion about virtual water inflow against total water flow within local area) are setup to reflect the trade frequency of products having high virtual water content and the importance of virtual water flow, respectively. The future fluctuations of every province are calculated using the precipitation and temperature output data from Beijing Climate Center Climate System Model in different climate change scenarios. The fluctuations are then used to estimate the variations about virtual water flows and relevant evaluation indexes. Results show that the virtual water flow intensity’s variations of the provinces with high intensity are much bigger than other administrative regions under climate change. It is also found that the virtual water flow have both positive and negative effects on the distribution of water resources, and the climate change has very small influence on the situation that most of the virtual water flowing in the developed areas come from water scare regions.