Because fire rescue personnel often enter unfamiliar buildings to perform critical tasks like rescues, the importance of finding new and improved ways to train route navigation is becoming paramount. This research was designed to compare three methods for training firefighters to navigate a rescue route in an unfamiliar building. Thirty firefighters from the Madison County, Alabama, area were trained to navigate through the Administrative Science Building at The University of Alabama in Huntsville. The firefighters, who had not had any experience with the Administrative Science Building prior to the experiment, were randomly assigned to one of three experimental training groups: Blueprint, Virtual Reality, or No Training (Control). After training, we measured the total navigation time and number of wrong turns exhibited by firefighters in the actual building. Participants were required to rescue a mock baby (a life-sized doll) following the specific trained route. Measures of test performance were compared among groups by using analyses of variance (ANOVAs). The results indicated that firefighters trained with virtual reality or blueprints performed a quicker and more accurate rescue than those without training. Furthermore, the speed and accuracy of rescue performance did not differ significantly between virtual reality and blueprint training groups. These results indicate that virtual reality training, if constructed and implemented properly, may provide an effective alternative to current navigation training methods. The results are discussed with regard to theories of transfer of training and human performance in virtual environments.