Abstract:As randomly occurring events, traffic accidents pose serious challenges to the collection of comprehensive data to understand how travellers respond to them and to quantify their impacts. The advent of mobile phones, with their wide spatial/temporal coverage and ubiquitous presence in metro areas, offers a new source of data to conduct such studies. In this study, the collision accident of Metro Line 10 in Shanghai, China on 27 September 2011 is carefully investigated based on data derived from anonymous mobile phone records. The evacuation process of the accident is studied, followed by an analysis of the impact of this accident on commuting in the city. After analysing 7 billion of mobile phone records for an 11-day period, the authors find that the evacuation follows a two-stage pattern. They then identify the commuters of Line 10 and study their commuting patterns in the day of accident and also in the subsequent days. They find that most of Line 10 commuters still preferred to use metro to complete their travels during the disruption period of Line 10, and returned to their typical commuting patterns immediately after Line 10 resumed service.