The increase in atmospheric carbon dioxide (CO2) concentrations due to emissions from fossil fuel combustion is contributing to recent climate change which is among the major challenges facing the world. Agroforestry systems can contribute to slowing down those increases and, thus, contribute to climate change mitigation. Agroforestry refers to the production of crop, livestock, and tree biomass on the same area of land. The soil organic carbon (SOC) pool, in particular, is the only terrestrial pool storing some carbon (C) for millennia which can be deliberately enhanced by agroforestry practices. Up to 2.2 Pg C (1 Pg = 10(15) g) may be sequestered above- and belowground over 50 years in agroforestry systems, but estimations on global land area occupied by agroforestry systems are particularly uncertain. Global areas under tree intercropping, multistrata systems, protective systems, silvopasture, and tree woodlots are estimated at 700, 100, 300, 450, and 50 Mha, respectively. The SOC storage in agroforestry systems is also uncertain and may amount up to 300 Mg C ha(-1) to 1 m depth. Here, we review and synthesize the current knowledge about SOC sequestration processes and their management in agroforestry systems. The main points are that (1) useful C sequestration in agroforestry systems for climate change mitigation must slow or even reverse the increase in atmospheric concentration of CO2 by storing some SOC for millennia, (2) soil disturbance must be minimized and tree species with a high root biomass-to-aboveground biomass ratio and/or nitrogen-fixing trees planted when SOC sequestration is among the objectives for establishing the agroforestry system, (3) sequestration rates and the processes contributing to the stabilization of SOC in agroforestry soils need additional data and research, (4) retrospective studies are often missing for rigorous determination of SOC and accurate evaluation of effects of different agroforestry practices on SOC sequestration in soil profiles, and (5) the long-term SOC storage is finite as it depends on the availability of binding sites, i.e., the soil's mineral composition and depth. Based on this improved knowledge, site-specific SOC sequestering agroforestry practices can then be developed.