Large areas of boreal forest in North America and Eurasia are frequently disturbed by wildfire. These fires alter ecosystem structure and function and affect climate through various biophysical and biogeochemical pathways. Fire-related forcings, however, are highly uncertain, can be opposite in sign, and depend on fire behavior as mediated by meteorology and intrinsic ecosystem properties. Our current understanding of large-scale fire dynamics is inadequate for fully characterizing their role in the climate system. This is particularly pertinent given the sensitivity of high latitudes and the large projected increases in fire frequencies during the 21st century. My dissertation aims to better characterize the controls on and feedbacks from boreal fires so that we may properly account for them in global change projections and potentially mitigate the impacts. I first quantified landscape-scale fire carbon emissions from a 2010 burn in Alaska using field measurements and fine-scale (30 m) remote sensing imagery. Accurate maps of fire emissions are needed to validate larger-scale models and quantify regional carbon fluxes, but are currently lacking due to spatial scaling issues. Here I show that by accounting for plot-level heterogeneity and species effects on spectral signatures, emission models can be generated from non-linear correlations between the differenced Normalized Burn Ratio (dNBR) and field data. Belowground combustion was quantified from soil cores and scaled to the site-level using spruce adventitious root heights. Species-specific allometric equations and visual estimates were used to characterize aboveground carbon losses. Results indicated that fire-wide combustion (1.98 ± 0.19 kg C m -2 ) was substantially lower than that in the core burning area (2.67 ± 0.24 kg C m-2 ) and sites (2.88 ± 0.23 kg C m-2 ) because of lower-severity patches and unburned islands. These areas constitute a significant fraction of burn perimeters in Alaska but are generally not accounted for in regional-scale estimates. This approach may be suitable for other fires in the region. In addition to the positive forcing from carbon emissions, forest fires in boreal North America exert a cooling effect due to relatively large increases in spring albedo from canopy destruction and tree fall. Although this forcing has been characterized at local and regional scales, its climate impacts have not been assessed. I simulated the continental-scale climate footprint of this cooling under various burning scenarios. Forest composition was characterized using a stochastic model of fire occurrence, historical fire data from national inventories, and succession trajectories derived from moderate-scale remote sensing (500 m). When coupled to an Earth system model, younger vegetation from increased burning cooled the high-latitude atmosphere, primarily in the winter and spring, with noticeable feedbacks from the ocean and sea ice. Results from multiple scenarios suggested that a doubling of burn area could cool the surface by 0.23 ± 0.09°C across boreal North America during winter and spring months (December through May). This has the potential to provide a negative feedback to winter warming across the domain on the order of 3-5% for a doubling, and 14-23% for a quadrupling, of burn area. Maximum cooling occurred in the areas of greatest burning and between February and April, reaching feedback potentials of up to 60%. Fire dynamics have been studied much less extensively in boreal Eurasia despite the region containing roughly 2/3rds of the world's boreal forests and displaying unique patterns of fire behavior. I used over a decade of satellite imagery to characterize variations in circumpolar fire behavior, immediate impacts, and longer-term responses. Compared to boreal North America, Eurasian fires were 58 ± 31% less likely to be crown fires, combusted 36 ± 5% less live vegetation, and caused 42 ± 5% less tree mortality. Eurasian fires also generated a 69 ± 9% smaller surface shortwave forcing during the initial post-fire decade, suggesting a near-neutral net climate forcing. Current global fire models were unable to capture the continental differences. I demonstrate that fire weather cannot explain the divergent fire dynamics and climate feedbacks. The primary drivers are shown to be species-level adaptations to fire, making this a preeminent example of species effects on continental-scale carbon and energy exchange.