AbstractGPS‐observed vertical ocean tide loading displacements show in Cornwall, southwest England, and in Brittany, northwest France, discrepancies of 2–3 mm with predicted values based on the isotropic Preliminary Reference Earth Model for the main tidal harmonic M2, yet in central Europe the agreement is better than 0.5 mm. By comparison of ocean tide models and validation with tide gauge observations, we demonstrate that the uncertainties in the former are too small to cause this disagreement. Furthermore, we find that different local models of the crust and different global elastic reference models derived from seismological observations can only reduce the observed discrepancies to 1–2 mm, which still exceeds the GPS observational uncertainty of 0.2–0.4 mm. It is customary to use the elastic properties of the Earth as given by seismic models. Previously, there has been insufficient evidence to determine how to modify these properties during the transformation from seismic to tidal frequencies to account for possible anelastic dispersion in the asthenosphere, and so this effect has been ignored. If we include this effect, then our discrepancies reduce further to 0.2–0.4 mm. This value is of the same order as the sum of the remaining errors due to uncertainties in the ocean tide models and in the GPS observations themselves. This research provides evidence in western Europe of a reduction of around 8–10% of the seismic shear modulus in the asthenosphere at tidal frequencies. In addition, we find that the asthenosphere absorption band frequencies can be represented by a constant quality factor Q.