The Federal Railroad Administration is interested in understanding the effect of warning reliability on motorist compliance to warnings at active grade crossings. The John A. Volpe National Transportation Systems Center conducted two studies to examine motorist behavior as a consequence of warning signal failure. Experiment one measured motorist behavior in response to false alarms (i.e., the presentation of a warning when no train was approaching). Experiment two examined how motorist responses to grade crossing warning signals were influenced by false alarms and missed signals (i.e., the failure of the warning system to signal an approaching train). The results of the experiments support the hypothesis that warning systems reliability exerts a predictable effect on motorists. As motorist’ perceive the warning system to be less reliable, they are more likely to violate the warning signal, perceiving little risk to their safety. Although it is not possible to define empirically the precise warning reliability required to achieve a desired level of compliance, measures for improving motorist behavior from an engineering and cognitive science perspective are considered and areas for future research are discussed.