The geology of the Boulonnais has been well studied since the early part of the last century [Gosselet and Bertaut, 1873; Olry, 1904; Pruvost and Delepine, 1921]. Extensive coal exploration added substantially to the general understanding of the geology of the region but as outcrop is poor, many questions remain. Gravity methods used in the analysis of geological structures have had a long and successful history in helping to study the earth's crust for scientific and applied objectives. Regional gravity data are particularly useful in mapping geographic distribution and configuration of density contrast of rocks. Previous gravity research shows the main trends of the structure. In most cases the regional Bouguer gravity hides the relationship between the geology and the shape of the anomaly caused by the perturbing body. New information can be obtained by filtering the maps. The purpose of filtering a map is to remove unwanted characteristics and enhance desirable characteristics that are diagnostic for the geology. Because of their simple mathematical forms, most potential field filters are in the spectral domain. It is advisable to transform the original unfiltered field to the spectral domain, apply the filter, then transform the filtered map back to the spatial domain for use in the interpretation. Several spectrally filtered versions of the original gravity map are used in this regional interpretation. In the case of the Boulonnais the most useful filters have been the horizontal component and the first vertical derivative. In the first instance computing the horizontal gradients of the gravity field permits us to localise the limit of the blocks and then the fault positions. The gravimetric field above a vertical contact of rock with different density shows a low on the side of the low density rocks and a high on the side of the high density rocks. The inflection point is located just on the contact of the two types of rocks. This contact can be outlined by locating the maxima of the horizontal gradient. In the case of a low dipping contact maxima stay close to the contact, but are displaced down dip. In the second instance the first vertical derivative acts as a booster for the short wavelength; this attenuates or destroys the effect of the regional field. The resulting map shows a better structure because in complex areas they give a better definition of the different bodies by separating their effects. In the case of the Boulonnais the first vertical derivative allows us to distinguish the depressed region from the uplifted one. The structural evolution of the Boulonnais-Artois area includes two main extensional events in the late Palaeozoic-early Cretaceous interval and an inversion in mid-late Palaeocene time. The new gravity data in combination with recent field and published data have provided a new insight into the structure of the Boulonnais-Artois area and a new interpretation is proposed. -- Fault patterns are oriented 110N and 040N in the Boulonnais and 140N in Artois areas. -- The linkage between the faults shows a relay geometry with transfer zones [cf. Morley et al., 1990 and Pea-cock and Sanderson, 1994]. The best example is located between Sangatte (near the tunnel) and Landrethun faults where overlapping synthetic faults with a relay ramp are imaged. -- There is no major continuous fault zone but a complex en echelon fault system. -- Linkage between Boulonnais and Artois fault is not well constrained. An important discontinuity between the two regions is apparent. This model underlines the importance of overlapping fault tips with the generation of transfer zones. These structures are also known in the Wessex and Weald basins [Stoneley, 1982; Chadwick, 1993] where heritage and inversion are significant.