Abstract: Understanding fire behavior is critical to effective tactical decision making on the fireground, particularly since fireground operations significantly impact the growth and spread of the fire. Computer-based simulation is a flexible, low-cost training methodology with proven success in fields such as pilot training, space, and military applications. Computer-based simulation may enhance fire behavior training and promote effective fireground decision making. This study evaluates the potential of the NIST Fire Dynamics Simulator (FDS) and Smokeview to be utilized as a part of a computer-based fire fighter trainer. Laboratory compartment fire experiments and full-scale fire experiments in a live-fire training facility were both conducted as part of the NIST Multiphase Study on Fire Fighter Safety and the Deployment of Resources. The laboratory experiments characterized the burning behavior of wood pallets to design a repeatable fire for use in the field experiments. The field experiments observed the effects of varying fire fighter deployment configurations on the performance times of fire fighter actions at a live fire training facility. These actions included opening the front door and fire suppression. Because the field experiments simulated numerous fire department responses to a repeatable fire, data were available to evaluate FDS simulation of heat and smoke spread, and changes in the thermal environment after the front door is opened and fire suppressed. In simulating the field experiments, the laboratory-measured heat release rate was used as an input. Given this assumption, this study has two objectives: 1) to determine if simulations accurately spread heat and smoke through a multi-level, multi-compartment live fire training facility 2) to determine if the simulations properly reproduce changes in the thermal environment that result from two typical fire fighter actions: opening the front door and fire suppression. In simulation, heat and smoke spread to measurement locations throughout the test structure at times closely matching experimentally measured times. Predictions of peak temperatures near the ceiling were within approximately 20% for all measurement locations. Hot gas layer temperature and depth were both predicted within 10% of the floor to ceiling height. After the front door was opened, temperature changes near the door at the highest and lowest measurement locations matched with temperature changes in the experiments. After fire suppression, FDS simulated temperature decay at a rate within the range measured in the field experiments and approximated the total rise of the hot gas layer interface in the burn compartment 250 seconds after suppression.