Abstract In arctic tundra, large and small mammalian herbivores have substantial impacts on the vegetation community and consequently can affect the magnitude of carbon cycling. However, herbivores are often absent from modern carbon cycle models, partly because relatively few field studies focus on herbivore impacts on carbon cycling. Our objectives were to quantify the impact of 21 years of large herbivore and large and small herbivore exclusion on carbon cycling during peak growing season in a dry heath tundra community. When herbivores were excluded, we observed a significantly greater leaf area index as well as greater vascular plant abundance. While we did not observe significant differences in deciduous dwarf shrub abundance across treatments, evergreen dwarf shrub abundance was greater where large and small herbivores were excluded. Both foliose and fruticose lichen abundance were higher in the large herbivore, but not the small and large herbivore exclosures. Net ecosystem exchange (NEE) likewise indicated the highest carbon uptake in the exclosure treatments and lowest uptake in the control (CT), suggesting that herbivory decreased the capacity of dry heath tundra to take up carbon. Moreover, our calculated NEE for average light and temperature conditions for July 2017, when our measurements were taken, indicated that the tundra was a carbon source in CT, but was a carbon sink in both exclosure treatments, indicating removal of grazing pressure can change the carbon balance of dry heath tundra. Collectively, these findings suggest that herbivore absence can lead to changes in plant community structure of dry heath tundra that in turn can increase its capacity to take up carbon.