What is already known:
What this study adds:
The effect of snow on roads constitutes a major threat to human health and safety, causing more fatalities in the US than all other types of high-impact weather (e.g. tornadoes, floods, lightning). The National Highway Traffic Safety Administration keeps a record of all automobile accidents resulting in fatalities. Included in this archive are a number of useful metadata parameters, such as the time of day, number of vehicles, the present weather observed by the reporting officer, etc. Herein, this metadata are examined as well as the prevailing synoptic forcing to better understand the factors leading to winter-storm-related road fatalities. A number of counterintuitive results were discovered. For example, fatalities do not occur more often in areas with infrequent snowfall or early in the season (i.e. in places and at times when drivers are less acclimated to snow). Rather, they have a democratic temporal and spatial distribution. While media attention is often focused on major pile-ups involving a large number of vehicles, the vast majority of fatalities involve just one or two vehicles. Despite the fact that primary and secondary roads are more likely to receive treatment and regular plowing, most fatalities (about 75%) occur on these types of roads. The meteorological conditions during crashes that occurred over two winter seasons are investigated. A large majority (about 77%) of the events happen within midlatitude cyclones or lake-effect snowbands. These are fairly predictable forms of weather, begging the question of what type of messaging was produced by the National Weather Service at the time of the accident. Only 45% of all accidents occur within a watch or advisory polygon, suggesting a more graduated approach for the warning/advisory system in winter storms, as opposed to binary polygons, has potential life-saving benefits.