Forest fragmentation is a pervasive consequence of human land use that creates novel forest boundaries in place of contiguous, intact forest. Boundary forests, or edges, experience environmental conditions distinct from the forest interior driven by lateral exposure to adjacent non-forest land cover. Forest edges tend to be hotter, drier and experience increased wind turbulence and atmospheric deposition with significant consequences for ecosystem processes and biogeochemical cycling. Much of what we know about forest edge structure and function derives from tropical forest research, despite prolific fragmentation in temperate forests. Building on recent field studies of temperate forest edges in the northeastern United States (US), I combine measurements from the US national forest inventory (NFI) with remotely-sensed maps of forest area to characterize broad patterns in the extent and impacts of fragmentation on temperate forest ecology. Using the US NFI to identify forest edges across a 20-state region, I report increased biomass and growth of edge forests compared their interior counterparts. I then compare the prevalence of forest edges in the US NFI and commonly-used forest maps to very-high-resolution land-cover maps, and I demonstrate that conventional methods of forest characterization systematically undercount and exclude forest edge area. Finally, I synthesize these findings to quantify aboveground carbon (C) cycling in New England using a novel approach that partitions forest C fluxes into forest edge and interior categories. I find that forest edges are disproportionately vulnerable to land-use conversion and are a critical component of both forest C uptake and emissions. Accounting for elevated growth rates in forest edges increases estimates of the net forest C sink in New England by 8.6% (4.36 Tg C). My dissertation research demonstrates the need to better understand the extent and effects of fragmentation in temperate forests, provides support for the treatment of forest edges as a distinct system, and highlights the need to include forest edges in current and future C accounting.