Land-use changes are the second largest source of human-induced greenhouse gas emission, mainly due to deforestation in the tropics and subtropics. CO2 emissions result from biomass and soil organic carbon (SOC) losses and may be offset with afforestation programs. However, the effect of land-use changes on SOC is poorly quantified due to insufficient data quality (only SOC concentrations and no SOC stocks, shallow sampling depth) and representativeness. In a global meta-analysis, 385 studies on land-use change in the tropics were explored to estimate the SOC stock changes for all major land-use change types. The highest SOC losses were caused by conversion of primary forest into cropland (−25%) and perennial crops (−30%) but forest conversion into grassland also reduced SOC stocks by 12%. Secondary forests stored less SOC than primary forests (−9%) underlining the importance of primary forests for C stores. SOC losses are partly reversible if agricultural land is afforested (+29%) or under cropland fallow (+32%) and with cropland conversion into grassland (+26%). Data on soil bulk density are critical in order to estimate SOC stock changes because (i) the bulk density changes with land-use and needs to be accounted for when calculating SOC stocks and (ii) soil sample mass has to be corrected for bulk density changes in order to compare land-use types on the same basis of soil mass. Without soil mass correction, land-use change effects would have been underestimated by 28%. Land-use change impact on SOC was not restricted to the surface soil, but relative changes were equally high in the subsoil, stressing the importance of sufficiently deep sampling.