An incised paleovalley offshore the estuary mouth of the Gironde River (Southwest France) was evidenced recently (Comptes Rendus de 1'Academie des Sciences de Paris 326 (1998) 701). Nevertheless its seaward extension to the shelf break was still unknown. New high-resolution geophysical data were acquired on the continental shelf in order to follow the incision until the shelf break. After processing the seismic lines, the shape of the valley was identified and the valley-fill in terms of sequence stratigraphy was interpreted.The valley-fill contains a single sequence corresponding to a fifth-order cycle (20 kyr). The three Systems Tracks described onshore by Alien and Posamentier (Journal of Sedimentary Petrology 63 (1993) 378; SEPM (Society for Sedimentary Geology) Special Publication, Tulsa (1994) 225) are present offshore. The Lowstand is poorly developed and is only present in the proximal part of the incised-valley. The Transgressive Systems Tract (TST) constitutes the bulk of the valley-fill, it is composed of landward-prograding estuarine clinoforms in the distal part of the valley (early TST, phase 1) and of landward-migrating tidal inlet-fill deposits in the proximal part of the valley (late TST, phase 2). The thin early Highstand Systems Tracts (HST) strata downlap seaward onto the maximum flooding surface and caps the TST. As expected in this wave- and tide-dominated environment, the tidal -and wave-ravinement surfaces, which bound the TST, are particularly well developed.The valley incision tapers 50 km offshore of the estuary mouth: the depth of the basal sequence boundary gradually decreases seaward and is locally truncated by the wave-ravinement surface around -70 m below the mean sea-level. These observations indicate that during the Lowstand times, rivers do not always generate continuous cross-shelf incised-valleys. They may adjust to a rapid drop in base level by modifying their channel morphology (i.e. deep large channels located upstream change downstream to shallow anastomosed channels with a weaker erosional ability). Therefore, the Lowstand fluvial channels will be poorly developed, and will eventually died out in the proximal part of the shelf because they are unable to keep pace with a phase of rapid sea-level drop. Similar observations have already been reported for the English Channel paleoriver (Geological Society Special Publication, London (1996) 203; These de Geologie Marine, Universite de Bordeaux 1 (1997) 265).