S U M M A R Y Two-ship multichannel seismic profiles, deep penetration (ECORS) and conventional seismic lines (LIGO surveys) are used to study the crustal structure of the Gulf of Lion (Western Mediterranean). 11 full ESPs (Expanded Spread Profiles) with total shot-receiver ranges up to 60 km were shot in 1981 perpendicular to the margin of the Gulf of Lion and in 1988 a deep MCS seismic profile (ECORS-CROP program) was performed parallel to the ESPs. These ESPs were analysed by matching traveltime and amplitude variations in both the x -f and z-p domains. The resulting P-wave velocity/depth model has the following features, (a) beneath the continental slope of the Provenqal margin a rapid rise of the Moho from 20 to 14 km and the existence of an anomalous 7.2-7.4 km s-' velocity layer, (b) from the base of the slope to the extensive salt-domes domain a 5-6 km thin crust which does not appear typically oceanic in nature, (c) quite typical oceanic crust up to the Sardinian margin. Gravity modelling is consistent with the seismic results. The OCB (ocean-continent boundary) could be placed north of that postulated by previous authors, where the data indicate a remarkably narrow transition between continental and 'oceanic' crust, or south where a typical oceanic crust, which correlates well with the domain of the salt domes and of large magnetic anomalies, has been determined.A very prominent reflector is clearly seen, at the base of the continental slope, on the seismic reflection profiles and corresponds to the top of an 7.2-7.4kms-' velocity layer. The high-velocity layer is 2-3 km thick where the crust is thinnest and has a limited lateral extent seawards. This anomalous crustal structure could be the result of extremely thinned and possibly broken up, continental crust underplated and intruded by partial melt, or could represent serpentinized peridotite material. Important questions about the evolution of the Gulf of Lion cannot be addressed using these new results alone without addition of other constraints. Nevertheless a two-stage mechanism of drifting and rifting of this part of the Western Mediterranean Sea is proposed.