— Gravity model used to model transport flow has proven invaluable in urban planning. Some recent work has proposed to use gravity model to predict user movements in cellular networks, however, there is generally a lack of practical evidence to justify the use of gravity model for such application. In this paper, we briefly summarize the theory of mobility gravity model which is derived based on the maximization of entropy. A simple algorithm which can be used to fit a set of observed trips as close to gravity model as possible is presented. We apply the algorithm to the census data collected in year 2000 from the train stations in the Tokyo metropolitan area. Statistical test is finally performed to validate the suitability in using gravity model to predict user movements for this set of data. Finally, how the model can be used to predict user behavioral change when there is a change in design parameters is discussed.