"It is recognized that local site conditions may affect the seismic ground motion, in terms of amplitude level, spectral content and time duration, thus increasing the seismic hazard level. Improving the way these site effects are accounted for is therefore a major concern, balanced however by economical constraints, which emphasize the need for inexpensive though reliable methods. In that context a seismic code regulation approach is often more appropriate than a specific study because of its simplicity and wide applicability. Indeed, the seismic codes consider site effects through site classification mainly defined with Vs30 (time-averaged velocity over 30 m) and associated spectra. However, despite Vs30 wide use as a proxy to site effects, it is often criticized as it does not seem fully relevant. An alternative approach, described in Cadet et al. (2007), proposes to characterize a site by Vsz (time-averaged over the top z meters) and f0 (fundamental resonance frequency). It also provides an empirical amplification function completely described by these two parameters. This approach was gauged and tested on a wide Japanese dataset, and it was proved to provide a larger variance reduction that Vs30 alone as f0 includes information on deep geology, while Vs30 carries only shallow information. The goal of this proposal is to deepen, extend and design a practical implementation procedure for this new approach, with both short term (site conditions, microzonation studies), and longer term (next generation of building codes) applications, with different steps: a) defining a protocol to characterize a site with Vsz and f0, or a direct use of the Rayleigh wave dispersion curve b)providing an estimation of the site amplification by updating the amplification function, including time duration elongation, non-linear and geometrical effects c)implementing this result in a regulation process by proposing a regulation-like shape of the analytical function"