The circulation dryer consists of drying, tempering, preheating and discharging stages, is the main equipment for rice drying in southern China. During circulation drying process, grain flows from top to bottom and returns to the top by elevator; air passes through the drying stages after being heated and carries away the vapor form grain; high temperature gas from boiler transfers heat to the air through the heating exchanger, and the flue gas passes through the heating tube in the preheating stage. Drying system may behave unexpectedly because of fluctuations of heat supply, medium flow, material flow and environmental state. In order to achieve greater drying capacity, milling quality, and energy efficiency, matching drying parameters with the system states dynamically is necessary. Due to the complexity of drying process, it is difficult to measure the system states accurately and comprehensively by sensors, therefore, it is meaningful to propose a mathematical model to predict the behavior of drying system. In this study, a transient mathematical models of various stages were developed for circulation counter-flow rice drying system based on heat and mass balance. The models consists of partial differential equations involving moisture content, grain temperature, humidity and air temperature. As the coupling of variables, analytic solutions of the PDEs could not be obtained, it was solved by finite difference method with first order upwind difference scheme. A algorithm was programmed on MATLAB and the distribution of moisture content, grain temperature, humidity and air temperature throughout drying bed varied with time was presented. The numerical simulation results showed that in counter-flow drying stages, moisture content decreased and grain temperature rise along the grain-flow, humidity increased and air temperature decreased along the air-flow; high moisture grain versus low temperature air, low moisture grain versus high temperature air, which accorded with demand of drying energy. When the operation condition was constant, moisture content alternated between descent-phase and plateau-phase with position in drying bed and time; grain temperature alternated between descent-phase and ascent-phase with position in drying bed; the maximum moisture point moved periodically in the first drying stage, and the minimum one was at the outlet of second drying stage constantly. Coefficient of variation(c.v.) of moisture distribution can be used to measure the uniformity of dried products, statistical results showed the c.v. of moisture distribution changed with time periodically, ranging from 0.006 to 0.059, and was minimum when drying system went through compete cycling time. At different operation conditions, the higher the inlet air temperature and initial moisture content, the higher the average drying rate, however, the effect of grain flow velocity on average drying rate was not significant. The model was validated by performing experiments in a circulation counter-flow rice dryer (5HP-20). The predictions of outlet moisture content and grain temperature vs. drying time were observed to be close to the measured values in the drying experiments, the root mean square error between predicted and measured values of moisture content and grain temperature were 0.99%d.b, and 0.49 ℃, respectively. The model can be applied to analyze and predict the circulation drying process.