Would you trust a robot to lead you to safety in an emergency? What design would best attract your attention in a smoke-filled environment? How should the robot behave to best increase your trust? Emergency systems such as exit signs and alarms have certainly improved survivability in an emergency but evacuations are typically so disorienting that individuals are still injured and killed, sometimes within arms length of an exit. In this research, we look at elements of human trust of robots in high-risk situations, such as emergency evacuation. Based on our initial studies, we have developed a system that builds immediate trust in the robot by making it behave and appear as a responsible authority figure and maintain this trust through interacting with the evacuee. This work is done in conjunction with Dr. Paul Robinette (GTRI) and Dr. Alan Wagner (previously at GTRI, now at PennState).