The objective of this research is to use testing and modeling to quantify realistic exposure fires under a railcar and compare these exposures to those in a fire resistance test. A series of tests were conducted on a railcar floor mockup, scaled to 40% of a full railcar width, exposed to fire from below. Two fires were considered, one representing a 1.9 MW diesel fire (e.g. resulting from a ruptured fuel tank) and another representing a 0.3 MW trash fire (e.g. resulting from a collection of trash and debris under the railcar). Two geometric configurations were tested including a floor with equipment box obstructions and a flat floor without undercar obstructions. For the diesel fire, the heat flux directly above the fire reached 75 kW/m 2 for the flat configuration and 95 kW/m 2 for the obstructed configuration, while gas temperatures directly above the fire reached 750°C and 950°C, respectively. Temperatures and heat flux varied greatly over the floor geometry for the realistic fires, resulting in thermal gradients that are not characteristic of a fire resistance test. Computational fluid dynamics simulations were used to model these different fire exposures under the railcar floor mockup as tested. The fire dynamics predicted were consistent with those measured. In the region of the mockup where the fire plume impinges, heat flux was predicted to within 11–22% of that measured. In the surrounding regions of the mockup, heat flux was predicted to within 22–40% of measured values. This level of agreement is appropriate for large-scale fire experiments, and the results demonstrate that the model is validated for use in the configurations considered in this study.