Abstract: Empty car adjustment model researched in this paper was part of Car Flow Estimation System. The car flow estimation model could be used to calculate the demand and availability of empty car precisely on different time periods. According to the dynamic change characteristics of empty car adjustment problem of railway network, a dynamic multi-stage empty car adjustment model was set up. The empty car stranded costs and unsatisfied demand penalty costs related to time factor were considered in the object function. The Heuristic Algorithm of simulated annealing was designed to solve the function. Finally, a simple network was taken to verify the model, and the results indicated that the model could handle the empty car adjustment problem under dynamic environment.