The number of collisions and fatalities at rail-highway intersections has declined significantly over the past thirty years, despite considerable increases in the volume of both rail and highway traffic. This paper disaggregates the improvement into its constituent causes. Negative binomial regressions are conducted on a pooled data set for 49 states from 1975 to 2001. The analysis concludes that about two-fifths of the decrease is due to factors such as reduced drunk driving and improved emergency medical response that have improved safety on all parts of the highway network. The installation of gates and/or flashing lights accounts for about a fifth of the reduction. The development of the “Operation Lifesaver” campaign, that seeks to inform the public on appropriate conduct at crossings, in the 1970s and early 1980s; and the installation of “ditch lights” on locomotives in the mid 1990s, each led to about a seventh of the reduction. Finally, about a tenth is due to closure of crossings resulting from line abandonments or consolidation of little-used crossings.