Based on the analyses of the existing problems in the forecast of passenger flow of urban rail transit in China,a forecast model framework and a method are proposed under the iterative feedback constraints. In the investigation,according to urban master planning and detailed regulatory planning along the railway,the basic data is processed to regain the balance between the population and the employment opportunities in the research by coordinating the development intensities of the whole and the local. Then,by taking into consideration the impact of vehicle flow on buses,the traditional four-stage model is improved to make motor vehicle flow assignment model,traffic distribution model and mode split model interact with each other and implement iterative feedback. As a result,the balance convergence is achieved. Finally,the proposed forecast model was testified based on the passenger flow in Suzhou Rail Transit Line 2. The results show that,when the iterative times for three particular years respectively reach 7,9 and 10,the convergence conditions and the accuracy requirement are both met.