We studied scattering versus intrinsic attenuation estimates in the vadose zone from a shallow VSP experiment conducted in the Lawrence Livermore National Laboratory (LLNL) facility. Using permanent downhole geophones and a vertical impact source, we estimated effective attenuation of the downgoing transmitted P-wave. We compared theoretical scattering attenuation estimates and finite-difference synthetics to the measured field values ( being a measure of attenuation). Using a selected range of impedance profiles of variance typical for a sedimentary basin, our estimates of are in the order of 20–85. Given the short propagation pathlengths involved (), we show that attenuation due to lateral heterogeneity is not significant. We analyzed additional distorting factors, including near-field presence, local impedance effects, and interference from reflections originating beneath receivers, and found that they may significantly impact attenuation measurement in near-surface studies, and result in biased values. From a comparison of the analyzed ranges of to the measured values, we deduced , which is consistently low, but whose inferred frequency dependence depends strongly on the scattering model assumed. The ranges for the factor resulting from scattering and distorting factors, and the intrinsic value were estimated as, respectively, 4–7, and 7–4. We identified two potential mechanisms which could lead to low values in the vadose zone: patchy saturation and squirt flow. We found through viscoelastic 3D synthetic modeling using a standard linear solid (SLS), that the field frequency dependence can be reproduced, although nonuniquely, for the studied range of impedance variance.