Fuel input layers for the FARSITE fire growth model were created for all lands in and around the Gila National Forest, New Mexico, using satellite imagery, terrain modeling, and biophysical simulation. FARSITE is a spatially explicit fire growth model used to predict the growth of wildland fires in terms of size, intensity, and spread. It requires eight data layers as input; fire behavior fuel model, crown closure, crown base height, stand height, crown bulk density, elevation, aspect, and slope. These input layers were created from a digital terrain model (elevation, aspect, and slope) and from base vegetation layers of biophysical settings, cover type, and structural stage using a methodology designed to be easily replicated by other fire management agencies. Biophysical settings describe long-term environmental conditions and this layer was created from a vegetation-based potential vegetation type classification modeled from hierarchical topographic rulebase terrain models. Cover type and structural stage layers were created from 1993 and 1997 satellite Thematic Mapper (TM) imagery of southwestern New Mexico. Fire behavior fuel models were assigned to each biophysical setting, cover type and structural stage category combination from an analysis of comprehensive field databases created by extensive plot sampling of the entire study area. An extensive accuracy assessment of the layers showed accuracy ranges from 25 to 87 percent for the potential vegetation type, cover type, and structural stage layers. Accuracy for the crown and surface fuels layers is between 40 to 70 percent. © University of Idaho 2000. Abstract reproduced by permission.