Fire behavior modeling systems are playing an increasingly important role in identifying areas of the wildland-urban interface (WUI) that could support intense and fast-moving wildfires. The modeling systems also can be used to prioritize areas for fuels reduction treatments. We used forest inventory data to create custom fire behavior fuel models for the Anchorage, Alaska, WUI-an area strongly impacted by a recent spruce bark beetle (Dendroctonus rufipennis) infestation. Eight custom fuel models were developed including a custom fuel model for a spruce bark beetle impacted forest type. NEXUS simulations indicate that the custom fuel models better describe forest structure and predict fire behavior than do parameterized standard fuel models previously used by local fire managers. Rate of spread and fireline ranged from 1-321 chains/hour and 1-2,549 Btu/ft per second, respectively, for the custom fuel models compared with 1-70 chains/hour and 1-7,929 Btu/ft per second, respectively, for the parameterized standard fuel models. Our study shows that it is both possible and feasible to create custom fuel models directly from fuels inventory data. This achievement has broad implications for land managers, particularly managers of the boreal forest, a region that is susceptible to wildfires but also home to a growing human population and increasing amounts of development.