In recent years, several studies have been conducted in the detection and observation of nonaqueous-phase liquids (NAPLs) in contaminated soils. Successful remediation of an NAPL-contaminated site requires appropriate characterization of both the plume extent and the soil volumetric NAPL content (θNAPL). Noninvasive geophysical techniques, such as time domain reflectometry (TDR), may be used to discriminate between θNAPL and soil volumetric water content (θw). Accordingly, the main aim of this work was to develop a TDR waveform interpretation method based on soil dielectric permittivity measurement and observation of the change in the reflected TDR signal amplitude at relatively long times from the waveform source. We demonstrated that the asymptotic value of the reflection signal coefficient can be univocally related with θNAPL in an unsaturated soil. In the procedure adopted, a new formulation of a dielectric mixing model was also derived, introducing a fourth phase that takes the NAPL presence into account. Multiple samples of sandy, silt loam, and loamy soils, at predetermined different θw and θNAPL, were tested, each test consisting in emitting a TDR signal to the soil sample and receiving and analyzing the reflected electromagnetic wave. An empirical dielectric mixing model was calibrated and implemented to estimate the θNAPL. Equipment calibration, measurement accuracy, and error sources, related both to the experimental procedure setup and to the sample preparation conditions, are discussed. The results show that the suggested methodology can be used to obtain predictions of volumetric NAPL content (θNAPL) with acceptable accuracy (R2 = 0.95).