In a highly reliable system, barriers can be used to reduce the risk of high consequence events. This paper identifies and evaluates barriers within a large safety model of the National Airspace System. The model contains thousands of nodes, so a heuristic procedure is proposed to automatically identify candidate barriers in the model. A case study of a single accident scenario is presented to show how the heuristic works and then subsequently how component elements of each barrier function are identified. An overall analysis is presented showing the distribution of barriers throughout the model and their overall effectiveness. No obvious trend is identified relating the number of barriers to the rarity of a given accident scenario.