||Shallow landslides often occur at depths less than 3 m. The shallow soil layer may be subject to continuous cycles of wetting and drying. Desiccation cracks may develop during the drying process. Other types of cracks may also develop. Soil cracks have a significant influence on the hydraulic and mechanical behavior of soil. Considering the importance of cracks in geotechnical and environmental fields, water infiltration into cracked soils is investigated in this thesis. Field and laboratory tests were conducted. Double-ring infiltration tests and two-stage borehole tests were conducted in the countryside of Zhenjiang. The tests were conducted at the same location but different depths to observe the change of seepage behavior and the wetting process with depth. A tracer test was also carried out to visualize the presence of preferential paths. Block samples were retrieved at locations next to each in-situ test. The desiccation crack network enhances the permeability such that the permeability of the in-situ silty clay becomes 1.0 x 10-6 ms-1, which is about 5 times of the usual permeability of the intact silty clay. Also, tensiometers recorded a fast advance of wetting front in the soil layer. A perched water layer was detected during the test. The tensiometer readings further illustrated the relative amount of flow in the horizontal and vertical directions at different soil depths. In the laboratory, a 3-directional permeability test device with a constant water head control system was successfully developed for characterizing permeability anisotropy. The device is able to accommodate a block sample 200 x 200 x 200 mm in size. Small diameter samples were taken from each block sample for standard one-dimensional permeability tests to demonstrate the effect of sample size. The results of the 3-directional permeability tests demonstrated that the anisotropy factor (kh/kz) increases from less than 1.0 at the ground surface to slightly larger than 1.0 at 1.0 m depth. The ratio of the two horizontal permeabilities is close to 1.0. The deviations of the large-sample tests are less than the deviations of the small-sample standard tests. Samples with a single empty or filled crack were tested under different boundary conditions. The cracks were formed in both Kaolinite samples and in dummy plastic samples. A theoretical method is suggested to predict the permeability of a single crack formed by rigid walls and filled with sand. The permeability along the rigid wall is about 6.0 times of that of the permeability of the sand fill.