A variety of ideas exist on how the 25 mm/yr motion along the North Anatolian Fault (NAF) is transferred into the Aegean domain. We have determined a model strain rate field from an inversion and interpolation of almost 300 GPS velocities from six different studies. In the process of obtaining a best fit between a model velocity field and the GPS velocities, no a priori assumptions about plate rigidity have been made. On the other hand, we have placed constraints on the distribution and style of zones of weakness within the region. These zones of weakness and their inferred style correspond to our knowledge of active faults in the region. We find that regional shear strain rates are extremely localised and equally distributed along the entire NAF \226 North Aegean Trough (NAT) system. Extensional strain rates are found to be negligible along this system. Although the zone of high strain rates terminates abruptly northeast of Evia, a zone of relatively high counterclockwise model rotation rates (around a vertical axis) continues into continental Greece. Next, we compare our surface strain rate field and inferred directions of no-length- change with fast polarisation directions in the upper mantle obtained from SKS wave splitting directions. There is no correlation between SKS directions and no-length- change directions. Within central and northern Greece there exists a strong correlation between extension and SKS directions, while off the western margin of Turkey extension directions are systematically directed more northward than the NE- SW SKS directions. These comparisons suggest that shearing along NAT is recent and has not involved large shearing at depth, and that the present-day crustal extension in the eastern Aegean is differently oriented from finite deformation in the lithospheric mantle. SKS directions in the northeastern Aegean Sea do however align with Miocene extension directions. It is therefore plausible that we observe a temporal change in overall extension direction from NE-SW to N-S. The long-period of diffuse NE-SW extension, typical for the Miocene, has caused measurable anisotropy at depth, but the present-day N-S extension is too recent to have (yet) done the same. We propose that the recent continuation of the NAF into the Aegean (<1 Ma) coincided with the change in extension direction.