Understanding the scaling effect on the spatial variation of soil organic carbon (SOC) is necessary for regional eco-environmental protection and for determination of reasonable soil sampling intervals. The objective of this study was to characterize the scaling effect on the spatial variation of SOC using wavelet multi-analysis techniques. This study was based on data from 223 typical soil profiles that were derived from the second national soil survey data from Guangdong Province. The soil profile data included soil types, soil organic matter (SOM) content, soil bulk density, soil depth and area. We estimated SOC density from depths of 20 cm and 100 cm in mountainous areas of Guangdong Province, and then developed an SOC density map with a grid size of 250 m by using the soil type map (1:200000). In detail, we selected four spatial sampling transects, which comprised two horizontal transects with a grid size of 250 m from 24° 50' N to 24° N latitude, and two left-oblique transects with a grid size of 250 m from 111° E to 114° E longitude, and from 115° E to 116° E longitude. These four transects reflect the general characteristics of the study area. By using the method of discrete wavelet transform (DWT) on the spatial sampling data of SOC density along the above four transects, we derived wavelet approximate information and wavelet detailed information over six different scales of 2×250 m (0.5 km), 2²×250 m (1 km), 2³×250 m (2 km), 2⁴×250 m (4 km), 2⁵×250 m (8 km) and 2⁶×250 m (16 km) respectively. The descriptive statistical characteristics of SOC density were tabulated, and then we calculated the variance of the wavelet detailed information of SOC density, which can then represent the magnitude of variation in SOC density. The results showed the following. (1) On the whole, the spatial heterogeneity of SOC density was strong and varied with different transects and soil depth, which was controlled by dominant influence factors on different scales. (2) The overall optimum scale of grid size on which SOC had the maximum spatial heterogeneity was equal to or greater than 1 km. (3) The variation of different scales of each transect varied with soil type and vegetation type, geomorphologic characteristics, human activities(e.g. land use, cultivation management) and other influencing factors. In summary, this study provided a reference for the application of discrete wavelet transform to the spatial variation of SOC density on large regional scales, and made the analysis of SOC variation patterns more convenient.