As studies on racial profiling and biased policing have begun to proliferate, researchers are debating which benchmark is most appropriate for comparison with police traffic stop data. Existing benchmark populations, which include populations estimated from census figures, licensed drivers, arrestees, reported crime suspects, and observed drivers and traffic violators, all have significant limitations. This article offers a new, alternative benchmark for police traffic stops, a benchmark that has not been previously applied or tested in a racial profiling research setting. The analysis presented compares traffic observation data, gathered at selected, high volume intersections during an ongoing racial profiling study in Miami-Dade County, Florida, to not-at-fault driver demographic data from two-vehicle crashes at those same intersections. Findings indicate that non-responsible drivers in two-vehicle crashes appear to represent a reasonably accurate estimate of the racial composition of drivers on the roadways at selected intersections and within areas of varying racial composition. The implications of this finding for racial profiling research are discussed, and suggested areas for future inquiry are identified.