Uplifts, with elevations of up to 2000 m and short wavelengths (30--35 km), flank three major grabens in central Greece: the Rio, Sparta, and Atalanti grabens. They are bordered on their landward sides by narrow basins oriented parallel to the graben axes: the Manesi and Trichonis basins at the Rio graben and Copais and Istiea basins at the Atalanti graben. The flexural origin of these uplift profiles is investigated here by using thin, broken plate models. It is demonstrated that the observed topography of the graben flanks is consistent with the upward deflection of elastic and viscoelastic plates in response to upward directed forces applied at the graben flanks. In order to evaluate to applicability of each model, their predictions are checked against the observations. The elastic predictive modeling fits well with the observed flexural wavelengths and the flanking seismicity. However, it fails to predict the graben widths and the inferred elastic layer thickness. In contrast, the viscoelastic model successfully explains the graben widths and the back basins as flexural hinterland basins and matches the seismological data and the time constraints of rifting. It is therefore suggested that flexural uplift with viscoelastic relaxation accounts, for the building of the graben flanks. The invoked viscoelastic models constrain the effective elastic thickness of the plates at 10 km for the Rio graben and 15 km for the Sparta and Atalanti grabens, suggesting a low-viscosity lower crust. Furthermore, they predict low rates of tectonic uplift of the order of 0.1 mm a-1 for the Atalanti graben, intermediate rates of 0.24--0.37 mm a-1 for the Rio graben, and high rates of 0.7--0.9 mm a-1 for the Sparta graben. The latter are quite possibly overestimated since napping events, capable of producing high local relief, are traced normal to the modeled profiles.¿ 1997 American Geophysical Union