This study aims to develop a model of passenger behavior when choosing trains from among various types of train service such as local and rapid services in urban railways, in order to evaluate passenger flows. This paper describes an online survey method for obtaining passenger activity data using departure information display images. A disaggregate demand model was developed for train choice behavior based on the survey data which were collected using the method. The developed model is put into practice by means of an application named "Train-Choice Model Viewer", which can calculate the time series data of passengers' volume at stations and in trains. It makes it possible to simulate the congestion level on each train and in each station when the time schedule is input.