Soil organic carbon (SoC) is considered to be a key index in evaluation of soil quality, soil degradation and soil C sequestration. In order to discuss the temporal-spatial dynamics of SoC in reversed desertification area, Yulin County is taken as a case. Data of SoC were obtained based on the Second National Soil Survey in 1982 and re-sampling in 2003. Methods to soil sampling, SoC determination, soil organic carbon density (SoCD) and storage (SoCS) calculation are soil auger, K2Cr3O7 oxidation and FeSO4 titration and area-weighted mean values, respectively. On average, SoCD and SoCS and their changes vary between different soil groups, soil C pools and sub-areas. During 1982 and 2003, soil organic carbon density (SoCS) for arable layer (0-20 cm), 1-m depth (0-100 cm) and whole profile (100±10 cm) increase 0.15 kgm-2, 0.29 kgm-2 and 0.31 kgm-2 respectively, being equal to increments of SoCS of 10.12 Gg, 19.06 Gg and 20.10 Gg, correspondingly. As for soil C pools, arable layer is much more sensitive to climate change, land use/management and other human activities than 1-m depth and whole profile. With regard to soil group and sub-area, the change of SoCD of sandy soil distributed in the northwest is much less than that of loessal soil distributed in the southeast, thus per unit area SoCS in the northwest is lower than the southeast. As far as soil genus, the SoCD of moving sandy soil and semi-fixed sandy soil is significantly lower than that fixed sandy soil, and sandy loamy loessal soil is less than light loamy loessal soil, which indicate that there is great potential to increase SoC in the study area. These data demonstrate that afforestation, agroforestry, tillage of soil and water conservation, and other soil restoration countermeasures can enhance soil C sequestration and mitigate the atmospheric concentration of CO2.