To investigate the effects of topsoil thickness and plant root-system type on the macropore structure and water infiltration of spoil–topsoil composite soils, an in situ soil-column experiment was conducted at a spoil disposal site. Three topsoil thickness treatments (10, 20, and 30 cm) were established for both Elymus dahuricus and Medicago sativa, with uncovered spoil serving as the control. CT scanning and infiltration tests were used to determine soil physicochemical properties, macropore characteristics, hydraulic conductivity, and infiltration characteristics. The Kostiakov, Horton, and Philip models were then applied to fit the measured infiltration data. The results showed that, with increasing topsoil thickness, soil nutrient contents generally increased and soil bulk density decreased under both plant treatments. Most CT-derived macropore indices also increased, whereas root length, root tip number, root branching point number, and infiltration capacity generally decreased. Root development and infiltration capacity were greater under E. dahuricus than under M. sativa, with the 10 cm topsoil treatment planted with E. dahuricus exhibiting the highest infiltration capacity. Near-saturated hydraulic conductivity ((K(h0)) was strongly associated with root characteristics (0.47 < r < 0.61, P < 0.01). All three models adequately described the infiltration process; however, the Horton model had the highest mean coefficient of determination and produced parameters with relatively reasonable physical meanings. Overall, appropriate combinations of topsoil thickness and plant species can effectively regulate macropore structure and water infiltration in reconstructed spoil soils, providing an important basis for ecological restoration and hydrological function recovery at spoil disposal sites.