Every year in the United States, accidents at highway-railroad grade crossings take place where the motorist hits the side of the train at night. In a portion of these nighttime accidents, the motorist fails to see the train in the grade crossing. One proposed solution to prevent such accidents is to mount retroreflective material on the sides of rail cars to make them more conspicuous. The objective of this research was to evaluate the effectiveness of several train-mounted reflector patterns to improve recognition of the train. Four patterns were evaluated in two experiments, using a human-in-the-loop driving simulator. In the first experiment, participants, located at a fixed position from a highway-railroad grade crossing and parallel roadway, viewed over a thousand scenes showing one of three items: a moving train, a moving motor vehicle, or nothing. The participants’ recognition of different reflector patterns was recorded and analyzed using the Receiver Operating Characteristic (ROC) based on Signal Detection Theory (SDT). In the second experiment, participants drove the simulator and reported all objects on the roadway, some of which were trains. The recognition distance from the participants’ position to the train was recorded and analyzed.