Abstract: Pore structure is the index reflecting the nature of the physical and mechanical properties of the soil. The resolution scale of the traditional medical CT is not high enough to identify the single grain or pore characterizing the meso-structure of geo-materials. An industrial micro-CT, with a high spatial resolution of 14 μm, is used to scan a Nanjing silty sand specimen fabricated by pluviation. The binary digital image sequence is acquired after preprocessing the CT initial images, and three-dimensional reconstruction is implemented. Arbitrary representative elementary volume (REV) is extracted, and three-dimensional binary matrix is used for the characterization of the REV and for the quantitative calculation. By analyzing the tomographic images, the volume porosity of the whole specimen is calculated, with an error of only 3.93% compared with the experimental porosity. The apparent porosity along the specimen fluctuates between 20.97 and 46.77%. Four typical sections, one horizontal, one tilting with an angle of 60 degrees to the horizon and two orthogonal vertical sections, are extracted from the REV at the bottom of the specimen. Statistical analysis of the preferred orientation is performed, and the results show that the minimum orientation angle occurs at the horizontal section, and one of the two orthogonal vertical sections exhibits the maximum orientation angle and the other exhibits the maximum anisotropic ratio, and the oblique section exhibits approximate isotropy, namely no obvious preferred orientation. The pore network model is used for spatial pore diameter analysis of three extracted REVs, and the results indicate that the maximum pore diameter distributes from 629 to 696 μm, and the minimum pore diameter does from 54 to 77 μm, and the average pore diameter does from 166 to 185 μm, and the pore diameter concentrates from 100 to 200 μm. The proposed method is capable of nondestructive quantitative analysis of pore structure variation of the spatial soil under any effect induced by heat, water or mechanics.