There is approximately 50 times more inorganic carbon in the global ocean than in the atmosphere. On time scales of decades to millions of years, the interaction between these two geophysical fluids determines atmospheric CO2 levels. During glacial periods, for example, the ocean serves as the major sink for atmospheric CO2, while during glacial–interglacial transitions, it is a source of CO2 to the atmosphere. The mechanisms responsible for determining the sign of the net exchange of CO2 between the ocean and the atmosphere remain unresolved. There is evidence that during glacial periods, phytoplankton primary productivity increased, leading to an enhanced sedimentation of particulate organic carbon into the ocean interior. The stimulation of primary production in glacial episodes can be correlated with increased inputs of nutrients limiting productivity, especially aeolian iron. Iron directly enhances primary production in high nutrient (nitrate and phosphate) regions of the ocean, of which the Southern Ocean is the most important. This trace element can also enhance nitrogen fixation, and thereby indirectly stimulate primary production throughout the low nutrient regions of the central ocean basins. While the export flux of organic carbon to the ocean interior was enhanced during glacial periods, this process does not fully account for the sequestration of atmospheric CO2. Heterotrophic oxidation of the newly formed organic carbon, forming weak acids, would have hydrolyzed CaCO3 in the sediments, increasing thereby oceanic alkalinity which, in turn, would have promoted the drawdown of atmospheric CO2. This latter mechanism is consistent with the stable carbon isotope pattern derived from air trapped in ice cores. The oceans have also played a major role as a sink for up to 30% of the anthropogenic CO2 produced during the industrial revolution. In large part this is due to CO2 solution in the surface ocean; however, some, poorly quantified fraction is a result of increased new production due to anthropogenic inputs of combined N, P and Fe. Based on ‘circulation as usual’, models predict that future anthropogenic CO2 inputs to the atmosphere will, in part, continue to be sequestered in the ocean. Human intervention (large-scale Fe fertilization; direct CO2 burial in the deep ocean) could increase carbon sequestration in the oceans, but could also result in unpredicted environmental perturbations. Changes in the oceanic thermohaline circulation as a result of global climate change would greatly alter the predictions of C sequestration that are possible on a ‘circulation as usual’ basis.