We use a dataset of road traffic accidents that were reported to police in Montreal between 2011 and 2022 to create a model that predicts the severity of a road traffic accident based on data about that accident, including characteristics like: the type of road where the accident took place, the weather, the time of day, etc. Our model powers an application that stakeholders can use to interactively predict the severity of a given accident. This tool could be used by 911 call takers to provide guidance about an appropriate emergency response, or by urban planners when making decisions about road planning.