The seismic codes of many countries base their site classifications on the Vs30 parameter, i.e., the average shear-wave velocity in the first 30 m of subsoil. And yet, despite the widespread use of this parameter, there is not universal agreement that this is a valid proxy for seismic amplification. There is no doubt that since the shear-wave subsoil velocity (Vs) is tied to soil rigidity, the profile of the former must have a role in site effects (Shearer and Orcutt 1987; Cranswick et al. 1990). However, seismic amplification appears too complex to be related to the Vs profile in the first 30 m alone. Harmsen (1997) showed largely scattered Vs30 site amplification plots and concluded that source directivity and topography also may play a role in site amplification. Hartzell et al. (2001) found a correlation between Vs30 and site amplification, but their conclusion is based on only four Vs30 values (measured only down to 16–17 m and extrapolated to 30 m), all falling in areas of little or no amplification. They did not perform a statistical correlation based on several data equally distributed among all site classes. Boore (2004) concluded that the National Earthquake Hazard Reduction Program (NEHRP) soil classes are a useful way of grouping sites according to site amplification. However, Boore's work involved only class C and D sites, with one exception. Wald and Mori (2000) and Mucciarelli and Gallipoli (2006) raised doubts about the validity of Vs30 as a proxy to soil amplification in tectonically active complex regions or in particular geological settings, such as in the presence of inversions in the Vs velocity profiles (Di Giacomo et al. 2005). Finally, Frankel et al. (2002) and Park and Hashash (2004) illustrated the role of amplification from the deeper (> 30 m) sediments. We re-analyze the empirical …