Abstract. Abstract. Presently, the lack of data on soil organic carbon (SOC) in relation to land-use types and biophysical characteristics prevents reliable estimates of carbon stocks in montane landscapes of mainland SE Asia. Our study, conducted in a 10,000-hectare landscape in Xishuangbanna, SW China, aimed at assessing the spatial variability in SOC and its relationships with land-use cover and key biophysical characteristics at multiple spatial scales. We sampled 27 one-hectare plots including 10 plots in mature forests, 11 plots in regenerating or highly disturbed forests, and six plots in open land including tea plantations or grasslands. We used a sampling design with a hierarchical structure. The landscape was first classified according to land-use types. Within each land-use type, sampling plots of 100 m × 100 m each were randomly selected, and within each plot we sampled nine subplots. This hierarchical sampling design allowed partitioning of the overall variance in SOC, vegetation, soil properties and topography that was accounted for by the variability among land-use types, among plots nested within land-use types, and within plots. SOC concentrations and stocks did not differ significantly across land-use types. The SOC stocks to a depth of 0.9 m were 177.6 ± 19.6 Mg C ha−1 in tea plantations, 199.5 ± 14.8 Mg C ha−1 in regenerating or highly disturbed forests, 228.6 ± 19.7 (SE) Mg C ha−1 in mature forests, and 236.2 ± 13.7 Mg C ha−1 in grasslands. In this montane landscape, variability within plots accounted for more than 50 % of the overall variance in SOC. The relationships between SOC, biophysical characteristics and land-use types varied across spatial scales. Variability in SOC within plots was determined by tree basal area, litter layer carbon stocks and slope. Variability in SOC among plots in open land was influenced by land-use type – SOC concentrations and stocks in grasslands were higher than in tea plantations. In forests, the variability in SOC among plots was related to elevation. The scale-dependent relationships between SOC and its controlling factors demonstrate that studies which aim to investigate the land-use effects on SOC need an appropriate sampling design reflecting the controlling factors of SOC so that land-use effects will not be masked by the variability between and within sampling plots.