In this paper geo-statistical theory and methods were used to study the spatial variation of soil water and the key impact factors in four landscape types in dry season in Karst cluster-peak-depression region, based on the grid sampling (10 m×10 m) in the permanent monitoring plots (200 m×40 m). A probe into the ecological process and mechanism of soil water was made through principal component analysis and correlation analysis. The results indicated that the soil water in dry season remarkably increased and coefficients of variation (CV) increased with the landscape transition from crops (Ⅰ) to manmade forest (Ⅱ), to secondary forest (Ⅲ), to primary forest (Ⅳ) along the descending gradient of disturbance in Karst cluster-peak-depression region. Good spatial autocorrelation existed in soil water in dry season in the four landscape types, among which negative and positive spatial autocorrelation distances reflected two different patches being. The radii of the latter half patches in Ⅰ, Ⅲ, and Ⅳ landscape types were approximately 50 m, where was just located in the boundary of the slope and the depression. The radium of the latter half patch in Ⅱ landscape type was about 75 m, where the location of the turning point reflected the transition of land uses. The spatial variation characteristics differed in the four landscape types. The semi-variance functions of soil water in Ⅰ, Ⅱ, Ⅲ and Ⅳ stages fit exponential, Gaussian, exponential and spherical models best, respectively. The sill (C0+C) and total spatial variance increased, while range decreased, along the descending gradient of disturbance. The values of \[C0/(C0+C)\] in Ⅰ and Ⅳ were 48.3% and 39.4%, respectively, which indicated that medium spatial correlation existed. While those in Ⅱ and Ⅲ were less than 25%, which indicated that strong spatial correlation existed. The Kriging contour maps showed the soil water in Ⅰand Ⅳ landscape types with convex distribution, Ⅱ with unimodal distribution, and Ⅲ with concave distribution. The results of primary component analysis suggested that elevation and the slope position were the key impact factors of soil water in the four landscape types in dry season. Aside from elevation and slope position, other key impact factors were different in the four landscape types. Moreover, even though the same factor in the four landscape types, details about functions and correlations differed through correlation analysis. Therefore, corresponding strategies of rational usage and management of water resources should be made according to the different key impact factors of spatial variation of soil water in the four typical landscape types in dry season in Karst cluster-peak-depression region.