FrostFire is a major field experiment and modeling effort to study the role of fire in boreal forests as a global change feedback and simultaneously provide fire managers with an improved capacity to predict fire severity based on meteorological conditions. The centerpiece of the experiment was the July 1999 ignition and continued observation of a boreal watershed near Fairbanks, Alaska. The experiment is being conducted on an LTER site by the USDA Forest Service Research, Pacific Northwest Station; the University of Alaska, Fairbanks; the Bureau of Land Management, Alaska Fire Service; and many other collaborators. In total, the experiment is studying and modeling three feedback mechanisms in relation to boreal fire: 1) changes in carbon stocks caused by immediate off-site transfer during the fire and by changes in decomposition and production following fire; 2) changes in energy budgets caused by fire; 3) changes in the structure of the ecosystem, with associated changes in carbon or energy budgets. Assessments of carbon emissions from fires involves a series of tasks that include the quantification of area burned, pre-burn biomass loadings, per-area biomass consumption, and emission factor (carbon release per unit of mass consumed). Substantial effort has been made in the FrostFire experiment to measure the distribution of biomass loading and consumption within the watershed; and classifying that variability into a scheme that can be used to assess biomass emissions elsewhere in the boreal forest. This paper presents our first results in measuring the extent of organic material consumption and vegetation mortality in four hundred hectares of a mostly black spruce forest burned in 1999. We are relating the measured consumption to the fuel moisture, forest characteristics, and meteorology using a methodology consistent with that we have developed for experimental fires elsewhere in temperate and tropical forests. The results will be used to characterize the change in carbon stocks during forest fires, and to relate the change to visually or remotely sensed features. Results of this kind should improve the prediction and inventory of carbon emissions from fires. Fire severity was measured in three ways: 1) intensive biomass inventories were completed before and after the burn at selected sites within the watershed, including the duff, moss, litter, woody fuels, herbs, shrubs, and trees; 2) extensive inventories were completed on a 200-meter grid on the entire watershed, including a vegetation survey and cursory inventory of biomass at each site; 3) vegetation and fire severity was mapped and classified from 1:6000 aerial photography flown before and after the burn. Data collected on the intensive sites and extensive inventory were used to assess the change in carbon pools in each cover type and fire severity class, and the photography was used to integrate the data across the watershed. A digital map of vegetation cover and fire severity was prepared.