<p>In mountainous and hilly areas, co-seismic landslides can considerably increase the damage caused by earthquakes. Although local shaking amplification can favor slope failures, the incorporation of site effects into regional scale seismic hazard assessment is hampered by the lack of quantitative data. Ideally, a reconnaissance level approach for wide-area detection of slopes susceptible to seismically-induced landsliding should consider such effects. As a step in this direction, we propose an approach that takes into account the contribution of litho-stratigraphic amplification to slope dynamic response. For this purpose, we exploit the results of ambient noise analysis, which are obtainable at low cost over wide areas. In this work, we focus on sites where litho-stratigraphic amplifications have been documented through long-term accelerometer monitoring. First, noise recordings are analyzed to invert the ellipticity of Rayleigh waves extracted from the noise in terms of a S-wave velocity vertical profile. Then amplification factors are estimated in terms of Arias Intensity through a simplified 1D site response modeling. Finally, to test the validity of our approach, the modeling results are compared with the amplification factors calculated from the accelerometer data. The test outcomes suggest that the proposed approach can provide estimates within a 50% approximation where slope materials lack strong anisotropies of mechanical properties. This proves to be a considerable improvement over the predictions of site responses based on a rough proxy such as V_S30.</p>