Spatial variability of soil organic carbon (SOC) of different land use patterns and soil types was examined in a county-wide red soil region of South China,using six sampling densities,14,34,68,130,255,and 525 samples designed by the method of grid sampling in 6 different grid sizes,labeled as D14,D34,D68,D130,D255,and D525,respectively.The results showed that the coefficients of variation (CVs) of SOC decreased gradually from 62.8% to 47.4% with the increase in soil sampling densities.The SOC CVs in the paddy field change slightly from 30.8% to 28.7%,while those of the dry farmland and forest land decreased remarkably from 58.1% to 48.7% and from 99.3% to 64.4%,respectively.The SOC CVs of the paddy soil change slightly,while those of red soil decreased remarkably from 82.8% to 63.9%.About 604,500,and 353 (P < 0.05) samples would be needed a number of years later if the SOC change was supposedly 1.52 g kg-1,based on the CVs of SOC acquired from the present sampling densities of D14,D68,and D525,respectively.Moreover,based on the same SOC change and the present time CVs at D255,the ratio of samples needed for paddy field,dry farmland,and forest land should be 1:0.81:3.33,while the actual corresponding ratio in an equal interval grid sampling was 1:0.74:0.46.These indicated that the sampling density had important effect on the detection of SOC variability in the county-wide region,the equal interval grid sampling was not efficient enough,and the respective CV of each land use or soil type should be fully considered when determining the sampling number in the future. Spatial variability of soil organic carbon (SOC) of different land use patterns and soil types was examined in a county-wide red soil region of South China, using six sampling densities, 14,34,68,130,255, and 525 samples designed by the method of grid sampling in 6 different grid sizes, labeled as D14, D34, D68, D130, D255, and D525, respectively. These results showed that coefficients of variation (CVs) of SOC decreased gradually from 62.8% to 47.4% with the increase in Soil sampling densities. The SOC CVs in the paddy field change slightly from 30.8% to 28.7%, while those of the dry farmland and forest land decreased remarkably from 58.1% to 48.7% and from 99.3% to 64.4% respectively.The SOC CVs of the paddy soil change slightly, while those of red soil decreased remarkably from 82.8% to 63.9% .About 604,500, and 353 (P <0.05) samples would be needed a number of years later if the SOC change was supposedly 1.52 g kg- 1, based on CVs of SOC acquired from the present sampling densities of D14, D68 , and D525, respectively. Based on the same SOC change and the present time CVs at D255, the ratio of samples needed for paddy field, dry farmland, and forest land should be 1: 0.81: 3.33, while the actual corresponding ratio in an equal interval grid sampling was 1: 0.74: 0.46.These indicates that the sampling density had important effect on the detection of SOC variability in the county-wide region, the equal interval grid sampling was not efficient enough, and the respective CV of each land use or soil type should be fully considered when determining the sampling number in the future.