Existing models of Jurassic and early Cretaceous relative motions between Antarctica and Africa significantly misrepresent the traces of fracture zones in the Riiser-Larsen Sea and Mozambique Basin. Improved visual fits of fracture zones, together with fits of magnetic reversal anomalies and other isochrons, provide a more reliable model. The model independently predicts fracture zone azimuths and magnetic isochron ages in the West Somali Basin, from which it is simplest to conclude that the West Somali Basin, Riiser-Larsen Sea and Mozambique Basin all formed simultaneously between the same two separating plates. This conclusion, together with the new rotations, supports the idea that Madagascar broke away from Antarctica at about chron M9, forming the west Enderby Basin. Combining data from all these basins in a joint inversion yields a set of finite rotations that more precisely describes the continuity of plate separation than the visually derived rotations do. The rotations can be used to generate a reconstruction that shows the Lebombo and Mateke-Sabi monoclines in Mozambique and the submarine Mozambique and Astrid ridges to be two sets of conjugate volcanic margins. Extrapolating the models extension rates back in time suggests that breakup and formation of these margins was closely associated in time with Karoo volcanism. The position of Madagascar in Gondwana determined with this reconstruction can be used as a constraint on the position of India in the supercontinent, which in turn may force a reassessment of the original locations of microcontinents such as Sri Lanka and Elan Bank.