Presented at the 2013 SMOS-Aquarius Science WorkshopOne of the challenges in collecting and comparing satellite and in situ data is the mismatch in their spatial coverage and the depth of sampling. In this study, we quantify how much of a difference is expected between in situ and satellite measurements of sea-surface salinity (SSS) in the presence of small-scale horizontal variability and near-surface vertical stratification. For this purpose, we make use of a solution from a global, eddy-resolving ocean data assimilation system. Sampling errors related to small-scale horizontal variability and vertical SSS gradients are typically small (<0.02 psu), but can be significantly larger locally (>0.1 psu), particularly near river mouths, boundary currents, parts of the ITCZ and in the Bay of Bengal. In such regions, sampling errors related to unresolved horizontal and vertical variability largely exceed in situ instrument noise and become an important source of uncertainty when comparing in situ and satellite data. The values and spatial patterns of the derived errors are discussed in the context of the Aquarius/SMOS overall error budget and can be used as a lower bound on expected differences between satellite and local estimates of SSS.