Multi-component seismic is increasingly being recognized not only as a solution to imaging problems, particularly where conventional seismic fails, but also as a tool for reservoir characterization. But geologists are often unenthusiastic to extract earth information from multi-mode seismic images because of their visual differences due to their own physical responses, and when the seismic images are equivalent the PS data does not provide any additional information by itself. The stratigraphic inversion is considered as a tool to transform seismic amplitude variations into petrophysical variations. With multicomponent acquisition, the AVA attributes of PP and PS modes are usually accessible, providing the additional shear wave information for a more constrained elastic inversion. Our inversion method simultaneously reconciles amplitude and time information from both PP and PS-wave data by equating the value of γ = V p /V s derived from the velocities or transit time ratio (γ t) and the value derived from the reflectivity ratio (γ a), providing not only the high-resolution of the V p /V s = γ r or the Poison ratio but also eventually V p , V s and densities (Garotta et al., 2000). A trace by trace inversion using simulated annealing process was described by (Dariu et al., 2003). In this paper we propose to examine some new possibilities and an extension of the method to 3D constraints in order to decrease the sensitivity to the noise and take into account the geological structure.