Core Ideas Computed tomography (CT)‐measured properties were in the order: soil under Q. liaotungensis > P. tabuliformis > C. korshinskii > M. sativa. The parameters measured by CT could help estimate saturated hydraulic conductivity (Ksat) in the Loess Plateau. Macropore quantity has a higher R2 for predicting Ksat than the morphology of macropores. The recovery of natural vegetation significantly improves soil macroporosity through root decay and biological activity. Macropores are preferential pathways for the movement of water to deep soil. However, the quantification of soil macropore structure and its relationship with soil hydraulic conductivity (Ksat) are not well understood. We characterized the macropores under different vegetation types to evaluate the effects of macropores on Ksat. Undisturbed soil cores were collected from four treatments: areas dominated by Quercus liaotungensis Koidz.(QLI), Pinus tabuliformis Carrière (PTA), Caragana korshinskii Kom. (CKO), and Medicago sativa L.(MSA). A medical computed tomography (CT) scanner was used to acquire images with a voxel size of 0.977 by 0.977 by 1.000 mm for depths of 0 to 360 mm. We used ImageJ software to quantify the macropore properties. Soil structure and Ksat improved with the succession of vegetation. The mean macropore volume fraction across the soil cores for the QLI, PTA, CKO, and MSA treatments were 6.6, 3.5, 1.3, and 0.6% within a 4900‐mm2 area, respectively. Macropore quantity (volume fraction and number) had a higher R2 for predicting Ksat than macropore morphology (branch density, connectivity density, and junction density). Moreover, the grayscale values were negatively correlated with Ksat and accounted for 78.8% of the variation in Ksat. Grayscale values, volume fraction, and the number of macropores were the best combination of CT‐measured parameters for predicting Ksat, accounting for 81.9% of the variation in Ksat. The CT‐measured parameters could be used to estimate Ksat in the upper root zone in the Loess Plateau.