Ground-penetrating radar (GPR) is used to track a dense non-aqueous phase liquid (DNAPL) injection in a laboratory sand tank. Before modeling, the GPR data provide a qualitative image of DNAPL saturation and movement. One-dimensional (1D) GPR modeling provides a quantitative interpretation of DNAPL volume within a given thickness during and after the injection. DNAPL saturation in sublayers of a specified thickness could not be quantified because calibration of the 1D GPR model is nonunique when both permittivity and depth of multiple layers are unknown. One-dimensional GPR modeling of the sand tank indicates geometric interferences in a small portion of the tank. These influences are removed from the interpretation using an alternate matching target. Two-dimensional (2D) GPR modeling provides a qualitative interpretation of the DNAPL distribution through pattern matching and tests for possible 2D influences that are not accounted for in the 1D GPR modeling. Accurate quantitative interpretation of DNAPL volumes using GPR modeling requires (1) identification of a suitable target that produces a strong reflection and is not subject to any geometric interference; (2) knowledge of the exact depth of that target; and (3) use of two-way radar-wave travel times through the medium to the target to determine the permittivity of the intervening material, which eliminates reliance on signal amplitude. With geologic conditions that are suitable for GPR surveys (i.e., shallow depths, low electrical conductivities, and a known reflective target), the procedures in this laboratory study can be adapted to a field site to delineate shallow DNAPL source zones.