This work provides new palinspastic palaeofacies reconstructions of SW Gondwana incorporating rotation of a Falkland/Malvinas microplate. We discuss the implications of this for the tectonic evolution of the southern South Atlantic and hence for the regional hydrocarbon potential. Existing Gondwana reconstructions display good fits of major continents but poorly constrained fits of microcontinents. In most continental reconstructions, the Falkland/Malvinas Plateau was assumed to be a rigid fragment of pre-Permian South American crust. However, it has been suggested, on the basis of palaeomagnetic data, that the Falkland/Malvinas Islands were rotated by ∼180° after 190 Ma. This rotation hypothesis has been successfully tested on the basis of Devonian stratigraphy and palaeontology, Permian stratigraphy and sedimentology and Late Palaeozoic and Early Mesozoic structure, making it unlikely that the plateau behaved as a rigid structure during breakup. We have explored the consequences of accepting this hypothesis for the tectonic evolution of SW Gondwana by compiling new palaeogeographic maps for the Permian–Cretaceous of the southern Atlantic area. To achieve a realistic close fit, we have devised a pre-rift proxy for the ocean–continent boundary for the South Atlantic. In order to produce the best fit, it is necessary to subdivide South America into four plates. The consequences of this are far-reaching. Our work suggests that although sedimentary basins were initiated at different times, three major tectonic phases can be recognised; in regional terms these can be thought of as pre-, syn- and post-rift. During the pre-rift time (until the Late Triassic), the area was dominated by compressional tectonism and formed part of the Gondwana foreland. The Falkland/Malvinas Islands lay east of Africa, the Falkland/Malvinas Plateau was ∼33% shorter and Patagonia was displaced east with respect to the rest of South America, in part along the line of the Gastre Fault System. Potential source facies are dominantly post-glacial black shales of Late Permian age deposited in lacustrine or hyposaline marine environments; these rocks would also be an effective regional seal. Sandstones deposited in the Late Permian would be dominantly volcaniclastic with poor reservoir qualities; Triassic sandstones tend to be more mature. There was significant extension from about 210 Ma (end-Triassic) until the South Atlantic opened at about 130 Ma (Early Cretaceous). In the early syn-rift phase, extension was accompanied by strike-slip faulting and block rotation; later extension was accompanied by extrusion of large volumes of lava. Early opening of the South Atlantic was oblique, which created basins at high angle to the trend of the ocean on the Argentine margin, and resulted in microplate rotation in NE Brazil. Intermittent physical barriers controlled deposition of Upper Jurassic–Cretaceous anoxic sediments during breakup; some of these mudrock units are effective seals with likely regional extent. During crustal reorganisation, clastic sediments changed from a uniform volcaniclastic provenance to local derivation, with variable reservoir quality. In the late rift and early post-rift phase, continental extension changed from oblique to normal and basins developed parallel to the continental margins of the South Atlantic. This change coincides with the main rifting in the Equatorial basins of Brazil and the early impact of the Santa Helena Plume. It resulted in widespread development of unconformities, the abandonment of the Recôncavo–Tucano–Jatoba rift and the end of NE Brazil plate rotation, which remained attached to South America. There was extensive deposition of evaporites, concentrated in (but not restricted to) the area north of the Rio Grande Rise/Walvis Ridge. Widespread deposits can be used to define potential regional elements of hydrocarbon systems and to provide a framework for relating more local elements. Our main conclusion is that the regional hydrocarbon potential of the southern South Atlantic has been constrained by the tectonic evolution.