Climate extremes, such as drought, may have immediate and potentially prolonged effects on carbon cycling. Grasslands store approximately one-third of all terrestrial carbon and may become carbon sources during droughts. However, the magnitude and duration of drought-induced disruptions to the carbon cycle, as well as the mechanisms responsible, remain poorly understood. Over the next century, global climate models predict an increase in two types of drought: chronic but subtle ‘press-droughts’, and shorter term but extreme ‘pulse-droughts’. Much of our current understanding of the ecological impacts of drought comes from experimental rainfall manipulations. These studies have been highly valuable, but are often short term and rarely quantify carbon feedbacks. To address this knowledge gap, we used the Community Land Model 4.0 to examine the individual and interactive effects of pulse- and press-droughts on carbon cycling in a mesic grassland of the US Great Plains. A series of modeling experiments were imposed by varying drought magnitude (precipitation amount) and interannual pattern (press- vs. pulse-droughts) to examine the effects on carbon storage and cycling at annual to century timescales. We present three main findings. First, a single-year pulse-drought had immediate and prolonged effects on carbon storage due to differential sensitivities of ecosystem respiration and gross primary production. Second, short-term pulse-droughts caused greater carbon loss than chronic press-droughts when total precipitation reductions over a 20-year period were equivalent. Third, combining pulse- and press-droughts had intermediate effects on carbon loss compared to the independent drought types, except at high drought levels. Overall, these results suggest that interannual drought pattern may be as important for carbon dynamics as drought magnitude and that extreme droughts may have long-lasting carbon feedbacks in grassland ecosystems.