Burn severity strongly influences post-fire vegetation succession, soil erosion, and wildlife populations in the fire-adapted boreal forest and tundra ecosystems of Alaska. Therefore, satellite-derived maps of burn severity in the remote Alaskan landscape are a useful tool in both fire and resource management practices. To assess satellite-derived measures of burn severity in Alaska we calculated the Normalized Burn Ratio (NBR) from pre- and post-fire Landsat TM/ETM+ data. We established 289 composite burn index (CBI) plots in or near four national park areas between 2001 and 2003 in order to compare ground-based measurements of burn severity with satellite-derived values of burn severity. Within the diverse vegetation types measured, a strong linear relationship between a differenced Normalized Burn Ratio (dNBR) and CBI for eight out of the nine fire assessments was found; R-2 values ranged from 0.45 to 0.88. The variations in severity among four pre- fire vegetation types were examined and a significant difference in the average dNBR and average CBI values among the vegetation types was found. Black spruce forests overall had the strongest relationship with dNBR, while the high severity white spruce forests had the poorest fit with dNBR. Deciduous forests and tall shrub plots had the lowest average remotely sensed burn severity (dNBR), but not the lowest ground severity among the vegetation types sampled. The tundra vegetation sampled had the lowest ground severity. Finally, a significant difference was detected between initial and extended assessments of dNBR in tundra vegetation types. The results indicated that the dNBR can be used as an effective means to map burn severity in boreal forest and tundra ecosystems for the climatic conditions and fire types that occurred in our study sites.