The extent and frequency of wildfires in Alaska’s boreal forest are predicted to increase in the coming century. In addition to natural sources of ignition, military lands in Interior Alaska are vulnerable to human ignitions due to their proximity to the road system and training activities. Recent wildfires—such as the 2013 Stuart Creek fire, which consumed nearly 365 km2 (90,000 acres), cost over $20 million to fight, and was sparked by an explosive ordinance on an army weapons range—demonstrate the need to test alternative fire management scenarios. One method that might reduce future large fires is to increase the level of fire suppression by changing the fire management planning options (FMPOs) for these areas from mostly Limited to Full protection. But will that method work well long-term? We used the Alaska Frame-based Ecosystem Code (ALFRESCO) vegetation-fire computer model to investigate how increasing fire suppression on military training lands could influence the extent and frequency of wildfire activity within the Upper Tanana Hydrologic Basin through the 21st century. ALFRESCO simulates wildfire, vegetation establishment, and succession—the dominant landscape-scale processes in boreal ecosystems in Alaska. We used a pair of climate models to bracket the uncertainty associated with projecting landscape changes. To simplify outputs, we focused on a single Representative Concentration Pathway (RCP) for greenhouse gases to drive the ALFRESCO model. Compared to the status quo (mostly Limited protection), changing all military lands within the study area to Full protection led to an increased number of fires, but a decrease in the total area burned, through 2100. These projected changes in fire regime also increased the amount of late successional coniferous forest present on the landscape. In contrast, keeping the areas in mostly Limited protection leads to more early successional deciduous forest on the landscape through the end of the century. The two climate models, however, drive the greater difference in results. Both models project future conditions warmer than today, but NCAR-CCSM4 projects a much warmer and drier future than MRI-CGCM3. Thus, ALFRESCO outputs using NCAR-CCSM4 predict greater fire activity and a declining ratio of coniferous to deciduous trees through the end of the 21st century. In contrast, ALFRESCO outputs using the MRI-CGCM3 model show an increase in the conifer:deciduous ratio over time. The effects of the alternative fire management planning options are subtle, so we recommend an economic study to determine if the cost of implementing such changes is warranted. Furthermore, we caution the results of this study are specific to a limited area within Interior Alaska. Future work will investigate whether modeling more large-scale fire suppression yields similar results.