Current accepted tools in risk assessment include techniques such as Fault and Event Tree analysis to predict the level of safety risk. To date, use of these tools at Grade Crossings has been limited and analysis has reverted to safety performance statistics to correlate results. There has been a missing link in modeling to be able to provide meaningful estimates of safety risk. Until now practitioners have been unable to use factors such as sighting, conspicuity, train speed, and human behavior characteristics within their numerical techniques. Working for Network Rail, the Railway operation company in the United Kingdom (UK) Halcrow has been using theoretical techniques to quantify safety risk at grade crossings. During this work we have applied a new tool in modeling safety risk, the Event Window. Using this concept, we have been able to relate variables such as sighting, user perception and predict human error against a timeline. The nearer a moving train to a grade crossing the less likely a user will cross in error. The advantage of this technique is that it takes into account the one common factor in past accidents, how users incorrectly judge train speed. In this paper we shall address the following questions: 􀂃 What is the benefit of quantifying the safety risk at a grade crossing? 􀂃 How do these risk assessments help organizations responsible for risk at grade crossings discharge their duties? 􀂃 What is the quality or usefulness of different types of evidence in building up these models? 􀂃 How can we improve on existing risk assessment techniques to make more informed and location specific decisions on grade crossing safety? 􀂃 How does the Event Window assist us in predicting safety risk? Tim Hess, AIMechE, Halcrow Group Ltd Jim Haile BSc CEng MIMechE, Halcrow Group Ltd