Spatial variation analysis of soil minerals is useful for ecological restoration and vegetation reconstruction. In this study, the spatial variation of soil minerals over a typical sloping farmland was investigated in a gorge karst region in southwestern China. The total study area (300 m × 200 m) was divided into grids of 20 m× 20 m and included 212 sample points. Soil minerals (SiO₂, Al₂O₃, Fe₂O₃, CaO, MgO, and MnO), in surface soil were measured. The spatial patterns of soil minerals were analyzed with classical statistics and geostatistics methods. The differences of the content of the six minerals were large, but the variation coefficients were small. The sum of SiO₂ and Al₂O₃ accounted for 85.99% of the sum of the six mineral components. SiO₂, Al₂O₃, MgO, and MnO, while Fe₂O₃ and CaO were normally distributed after square and reciprocal transformed, respectively. SiO₂, Al₂O₃, Fe₂O₃, CaO, MgO, and MnO were best fitted by an Exponential model, with high coefficients (R²) or low residual sum of squares (RSS) indicating that the fitted models could reflect the spatial variation of soil minerals. The nugget (C₀) was low (3.0%-43.6%), indicating that the soil minerals were strongly autocorrelated over the study region, and that their spatial patterns were influenced by structural factors. These spatial patterns varied over a small range (25.5-210.3 m). The spatial ranges of SiO₂ and MnO were large and similar (210.3 m and 195.9 m, respectively), and other were relatively small with strong spatial dependence. On Kriging contour maps, SiO₂ and MnO had similar spatial pattern of high values on the slope top and low values on the bottom slope, and were affected by slope position and anthropogenic disturbances; the patches of the Fe₂O₃, CaO and MnO were relatively homogeneous, and were mainly influenced by topography. Al₂O₃ posed a unimodal distribution, increasing along elevation ascending. Therefore, reducing disturbance and increasing vegetation would play good roles in protecting and regulating soil minerals.