Abstract: Origins of passenger flow of subway stations lays the foundation for planning of rail networks,optimizing connections between rail transit and ground transportations and setting up of the related facilities.Based on numerous signaling data of mobile phones with location information,a method for investigating origins of passenger flow of subway stations is proposed.On the basis of identifying distribution of Origin-Destination (OD) of passengers throughout a day,considering the fact that signal of mobile phones would be switched to different location areas (LA) when entering or exiting subway stations,time and location of a trip can be then justified.Comparing the time when passengers entered subway station with the departure time of all trips throughout a day,a complete trip chain of passengers who taking subway can be obtained.Then origin areas of these trips can be computed.Three typical stations of Beijing Subway Line 2 are selected for case studies.The computed results are compared with the data from Auto Fare Collection (AFC) machines in subway stations and the results of questionnaires.The comparisons show that the proposed algorithm is able to correctly identify more than 65 % of station demands,and the average error of identifying origins of passenger flow is about 5%.Finally,the distribution characteristics of the sources of subway passengers are analyzed.