Recently the evaluation of the travel time reliability has become great concern in transportation planning and many studies have investigated the travel time reliability. However most of them have focused on the reliability regarding road transports. Then the reliability of rail transport is focused on in this study. Authors have already investigated the departure time decision of railway users in Tokyo Metropolitan area (TMA) and developed a model describing a spare time for not being late behind the desired arrival time by considering the heterogeneity of the distribution of the late arrival. In order to estimate the coefficients of the model, we set a strong condition that the railway users in TMA recognized that the delay of arrival time
obeyed exponential distribution. However the difference between the desired arrival time and the actual arrival time does not always obey exponential distribution. Then, we started the research to ease this strong condition by adopting the mixture distribution model. There
are few studies investigating departure time decision dealing the mixture model. An internet survey was conducted to collect the data for the study. The question items are name of origin and destination stations, travel path, departure time, desired arrival time, required travel time, distribution of arrival time and socio-economic attributes.

urban railway, reliability of travel time, mixture distribution model