Based on the System Dynamics approach, a forecasting model for China's ration demand, named SDCRD, is developed. To further improve the fitness for historical data of traditional System Dynamics model, Support Vector Machine (SVM) technique is also applied, which has the ability to increase the efficiency and accuracy of SDCRD. Considering the ration demand system and other influential factors, such as the urbanization rate, residents' consumption habits, per capita income, and scientific-technical progress, etc. as a holistic system, SDCRD makes a more objective and accurate forecast for China's future ration demand. The prediction results indicate that the amount of overall ration demand in China has a moderate increase from 228.935 million tons in 2011 up to 230.65 million tons in 2020, with an average annual growth rate of 0.075%. Furthermore, the proportion of urban ration demand to total consumption steadily increases by years, from 36.77% in 2011 to 43.02% in 2020, whereas the rural ration demand descends gently from 144.754 million tons in 2011 to 134.062 million tons in 2020, which reflects the trend of structural changes of China's ration demand in coming decade.