In the Tokyo Metropolitan Area (TMA), train congestions of several railways still have been very severe during morning rush hour though railway operators have increased their train capacities to alleviate the peak-period congestion. To evaluate the policies such peak-spreading by travel time demand management and the introduction of flexible work-start time systems, it is necessary to construct appropriate departure time choice model of rail passengers. Though various types of scheduling model are proposed and tested in existing literature, it is still not clear what kind of model specification would be more suitable for the complex urban rail commuting patterns. In this study, we firstly carry out a simple regression analysis between passengers’ arrival time to understand relationships between departure time and trip distance. Then, we estimate and compare some departure time choice models and examine characteristics of each railway passenger in the TMA.