Ground penetrating radar (GPR) data were explored during a physical model experiment that utilized a polyethylene tank and an abiotic sand matrix to simulate water contamination in the vadose zone under fluctuating water table conditions. The main objective of this experiment was to determine the source of GPR reflections from a changing water table in a controlled experimental environment. During the experiment, the water table was varied by injecting water into the bottom of the tank and subsequently draining it. The experiment was conducted over a number of days, during which various conditions (unsaturated, saturated, residual saturation) were characterized using GPR measurements. As a result of this approach, a more comprehensive comparison was possible for each step and results were compared with mass balance information. Results of the study indicate that, under saturated conditions, the main reflector of GPR energy is indicative of the capillary fringe and not the actual water table. Well readings and estimates of the mass balance of water input into the system confirm this interpretation.