Non-invasive geophysical techniques are increasingly used to study the unsaturated zone. In particular, cross-hole methods can are able to infer more detailed information about the subsoil than surface measurements. Two borehole Ground Penetrating Radar (GPR) techniques are discussed in our contribution: Zero-Offset Profiling (ZOP) and Vertical Radar Profiling (VRP). We make a direct comparison of these methods in a field case (Trecate site, Northern Italy), to explore each method’s capabilities and limitations. Our analysis is focused on the results in the vadose zone and shows that the dielectric relative permittivity profiles recovered from ZOP and VRP first-break inversions are in strong disagreement, providing very different permittivity profiles. The analysis of synthetic radargrams shows the presence of an electromagnetic (EM) wave established by the joint presence of the air-filled borehole within a higher permittivity surrounding soil. This event has a velocity intermediate between the soil and air speed values, and interferes with the picking of first arrivals in the VRP mode. The numerical simulations are performed with different borehole diameters, confirming that the velocity of the first recorded event depends on the ratio between the wave length in air and the finite dimension of the borehole. Once these arrivals in the simulated VRP radargrams are recognized, their contribution can be removed by picking the “direct”ù arrivals, that correspond to the waves that directly propagates from source to receiver, through the unsaturated zone. Once the borehole effects are accounted for, the comparison between the ZOP and VRP permittivity profiles is reasonable and reveals the different resolution of these techniques, focusing on the information that can be inferred for hydrological characterizations. Thus, VRP surveys in vadose zone must be accurately interpreted, as the electromagnetic waves may propagate via guided modes along the borehole. Neglecting this phenomenon might generate misleading estimations of geophysical properties and the subsequently translation in hydrological quantities could leads to erroneous physical states; a very relevant issue in hydrological model calibration.