Knowledge about the annual and seasonal patterns of organic and inorganic carbon (C) exports from the major rivers of the world to the coastal ocean is essential for our understanding and potential management of the global C budget so as to limit anthropogenic modification of global climate. Unfortunately our predictive understanding of what controls the timing, magnitude, and quality of C export is still rudimentary. Here we use a process-based coupled hydrologic/ecosystem biogeochemistry model (the Dynamic Land Ecosystem Model) to examine how climate variability and extreme events, changing land use, and atmospheric chemistry have affected the annual and seasonal patterns of C exports from the Mississippi River basin to the Gulf of Mexico. Our process-based simulations estimate that the average annual exports of dissolved organic C (DOC), particulate organic C (POC), and dissolved inorganic C (DIC) in the 2000s were 2.6 ± 0.4 Tg C yr−1, 3.4 ± 0.3 Tg C yr−1, and 18.8 ± 3.4 Tg C yr−1, respectively. Although land use change was the most important agent of change in C export over the past century, climate variability and extreme events (such as flooding and drought) were primarily responsible for seasonal and interannual variations in C export from the basin. The maximum seasonal export of DIC occurred in summer while for DOC and POC the maximum occurred in winter. Relative to the 10 year average (2001–2010), our modeling analysis indicates that the years of maximal and minimal C export cooccurred with wet and dry years (2008: 32% above average and 2006: 32% below average). Given Intergovernmental Panel on Climate Change-predicted changes in climate variability and the severity of rain events and droughts of wet and dry years for the remainder of the 21st century, our modeling results suggest major changes in the riverine link between the terrestrial and oceanic realms, which are likely to have a major impact on C delivery to the coastal ocean.