The spat ial patterns of soil organic carbon (SOC) are closely related to changes in the global climate In order to quantify spatial patterns of SOC density in Hebei Province, China, three different methods, i e mult iple linear regression (MLR), universal kriging (UK) and regression-kriging (RK), coupled with auxiliary topographic factors extracted from a 1:250 000 DEM (cell size is 100 m) were applied to predict spatial patterns of SOC density for Hebei Province The results show that the sum squared error (SSE) of the MLR method was quite large with only 18 6% of the total variation explainable, the UK method lowered SSE but widened the range of SOC density as compared with the MLR method However, it can explain 53% of the total variation and detect local variation of lower SOC density in southeast of Hebei Province When the RK method was applied, the SSE decreased significantly It not only explained 65% of the total variation, but also better reflected the relationship between SOC density and landform and local variation, indicating that it is the best one for predicting spatial patterns of SOC density.