Present-day active tectonics in the Eastern Mediterranean relates to the rapid west to southwest escape of Anatolia and Aegea with respect to Eurasia. This motion is accommodated along two major right-lateral faults: the North Anatolian Fault (NAF) in Turkey and the Kephalonia Fault (KF) in Greece. Although kinematics impose that a zone of right-lateral shear stretches from the NAF to the KF, surface deformation studies indicate that none of these two faults cuts through continental Greece, and apparently do not connect. Westward propagation of the NAF is thought to occur presently in the pull-apart regime of the North Aegean Trough (NAT) whereas active N-S opening occurs in the Corinth Rift (CR) east of the KF. Continental Greece appears as a broad (>200km) region of active N-S extension. Based on a continuous strain rate and velocity field (50 km resolution) for the entire Mediterranean plate boundary zone, we investigate where and how deformation is distributed within this region and propose that the complex surface tectonics actually reflects a much simpler shear regime already at work in the mantle. The new model combines a least-squares fit to GPS velocities from multiple studies on land with a detailed data set of the location and style of active faults both on land and offshore. The first-order feature is the identification of a large rigid Aegean block covering not only the Aegean Sea but also the continuation of the continental backstop towards the deep Mediterranean basins. Right-lateral shearing along the NAF and the KF amounts to 22 mm/yr and 25 mm/yr relative motion respectively, whereas opening in the CR does not exceed 14 mm/yr, leaving about 11 mm/yr of distributed extension in northern Greece. Extension directions along the CR, in northern Greece as well as in NW Turkey (south of Marmara Sea) are all NNE-SSW, normal to existing grabens. The most striking result is that the narrow bands of high model rotation rates (around a vertical axis) associated with the NAF and the KF propagate into continental Greece well beyond their surface traces. Our results suggest that surface block rotation in this area is compatible with localized shearing at depth. Spatial wavelengths in the deformation pattern ($>$150km) further suggest that this deformation involves not only the crust but the mantle as well. The age of the propagation of the NAF into the Sea of Marmara (0.2 Myr) and the age of the Corinth Gulf and Kephalonia Fault ($<$1Myr) indicate that the present-day pattern was established very recently. We thus propose that the entire area between the NAF and the KF is a nascent pull-apart zone between two lithospheric-scale transcurrent faults, in response to established shear zones in the mantle. We test our new kinematic model by comparing results with fast-polarization directions of SKS waves, stretching of metamorphic lineations, and paleomagnetic measurements.