Soil organic carbon (SOC) content and its spatial heterogeneity are important components of regional and global carbon sequestration. However, the spatial variability of SOC in karst areas is poorly understood, especially in southwest China in areas of severe land degradation. In the present study, we selected isolated areas in the closest possible proximity to cultivated slopes in order to ensure similar soil and microclimate conditions. The spatial heterogeneity of cultivated slopes under long-term cultivation and abandoned lands under natural restoration in a karst peak-cluster depression area were analyzed by using the grid sampling method at 10 m × 10 m. A total of 230 soil samples were collected, including 123 from the cultivated slopes and 107 from abandoned lands. We used a one-way analysis of variance (ANOVA) to evaluate the differences between the plots. The results showed that the average SOC content in the abandoned lands was ~5 times greater than that in the cultivated slopes, which indicates that vegetation recovery significantly improved the SOC content. The estimate of the spatial pattern and scaling of the studied SOC was performed by geostatistical analyses using semivariograms to determine the average variance between samples collected at increasing distances. The results showed that all semivariograms of the two plots were fitted to an exponential model. The sill values of the semivariogram functions for the abandoned lands were about 15 times greater those of the cultivated slopes, which indicates that the spatial variation of SOC in the abandoned lands was higher than that in the cultivated slopes. Stepwise regression and redundancy analysis showed that factors influencing the spatial distributions of SOC in the cultivated slopes and abandoned lands were different. In particular, land cover type, slope position, rock ratio, and their interaction explained 9.1%, 6.3%, 4.6%, and 17.0% of the total variance of SOC in the cultivated slopes, respectively. Soil water content, slope gradient, rock ratio, and their interaction explained 26.0%, 10.7%, 7.2%, and 3.6% of the total variance of SOC in the abandoned lands, respectively. We conclude that land abandonment increased both the SOC content and its spatial heterogeneity in karst areas in southwest China. In addition, the degree of human disturbance and soil water content are the primary factors in determining the spatial variation of SOC in the cultivated slopes and abandoned lands, respectively. Moreover, the rock ratio is an important factor for controlling the spatial variation of SOC in both the cultivated slopes and abandoned lands. Our findings can provide a better understanding of the relationships between ecosystem restoration and SOC recovery and can give scientific basis for karst ecosystem restoration.