A variable-rate infiltration experiment was conducted in a sandbox to demonstrate that distinctive patterns are produced in transient ground-penetrating-radar (GPR) data collected during wetting and drying events. The observed GPR response was found to be very consistent with the results of numerical simulations performed using finite-difference time-domain modeling of GPR coupled with a 1D unsaturated flow model (HYDRUS-1D) for which the sand hydraulic properties were determined independently using core samples. Despite this agreement, few methods are available that can efficiently analyze transient GPR data to make a quantitative link between observed responses and the hydraulic properties of soils. To address this problem, a computationally efficient method is proposed that is analogous to coherency analysis used in multioffset surveys. The new method isbased on the calculation of semblance along trajectories through transient GPR data. Each trajectory represents a specific GPR arrival, e.g., the ground wave and reflections from the wetting front and subsurface boundaries. The specific path of the trajectories is controlled by the hydraulic properties of the soil, just as the normal-moveout trajectories used to calculate semblance in multioffset data are controlled by wave velocity. Because the method is based on the output of 1D unsaturated flow models, it can be used for situations with complex hydrologic boundary conditions. Good agreement was found in this study between the calculated trajectories and the arrivals observed for both simulated and empirical GPR data. A sensitivity analysis performed in this study suggests that most parameters of the Mualem–van Genuchten soil model can be identified using this approach to coherency analysis of transient GPR data.