In the Pyrenees, a phase of bulk vertical shortening (D-3) is indicated by a horizontal crenulation cleavage (S-3), which is developed in a pre-existing schistosity (S-2). Porphyroblasts of the Lys-Caillaouas massif regionally preserve: (1) the initial pre-D-3 orientation of S-2 in the form of a vertical and E-W-trending preferred orientation of their inclusion trails; and (2) a vertical L(2) stretching lineation defined by the elongation of the inclusions. These data are consistent with N-S compression and crustal thickening during D-2. The angle between the inclusion trails (S-2,S-i) and the (external) schistosity (S-2,S-e) records the subsequent rotation of S-2 from Vertical to gently dipping orientations. At higher crustal levels S-2 remained steep (suprastructure) and little D-3 deformation is evidenced. With depth, D-3 deformation progressively intensifies and zones can be distinguished with opposite dip directions of S-2, opposite apparent porphyroblast ''rotation'' senses and opposite F-3 crenulation asymmetries. These (micro)structures on a N-S section through the Lys-Caillaouas and the Garonne Dome massifs record a regional partitioning of bulk coaxial vertical shortening. This deformation partitioning was probably controlled by a not-exposed gneiss dome, which behaved relatively rigid as compared to its metasedimentary cover. In zones of non-coaxial D-3 deformation, S-2 was extended, reactivated and forms the dominant schistosity. Where D-3 was locally more coaxial, S-2 was shortened, isoclinally crenulated and S-3 forms the macroscopic cleavage. The fact that D-3 extension in the Lys-Caillaouas massif immediately followed D-2 compression and that similar extension is recorded in large parts of the Variscan foldbelt, suggests gravitational collapse of a thickened orogen. This is consistent with the flattening type of D-3 strain recognized in a number of Pyrenean Variscan massifs. Peak high-T, low-P metamorphic conditions coincide with the onset of orogenic extension, when very little crustal thinning could have been accomplished yet. This suggests that heat-induced strain softening within the orogen may have triggered its collapse.