We study the impact of travel-time reliability on trips made by railway passengers. Unlike most of the studies in this area, which make use of stated preference survey data, we use a revealed preference data set obtained by measuring the railway reliability and the number of season-ticket holders on the Dutch railway network. We make use of six travel-time reliability indicators, including the standard deviation and the 80th minus the 50th percentile of travel time. Our results indicate that the 80th minus the 50th percentile indicator best explains the fluctuations in the number of season-ticket holders. A 10% improvement of this indicator results in a 1.47% increase in the number of season-ticket holders. © 2010 Elsevier Ltd.