Existing models of relative motion between the South America and Africa plates significantly misrepresent the azimuth of Lower Cretaceous seafloor spreading in the South Atlantic Ocean. An improved model is derived from fits of fracture zones, magnetic reversal anomalies, and the edges of a rifted large igneous province at the Northeast Georgia Rise and Agulhas Plateau. An absolute date for this rifting event is not known but, by assuming least change in spreading rates during magnetic chron C34, rifting can be estimated to have occurred at ~100 Ma. This modelling demands a plate divergence history that involves diachronous opening of the South Atlantic, consistent with published estimates of the ages of breakup from sedimentary basins on the South American and African passive margins. The diachronous opening lasts approximately 40 million years, during which time it must be accommodated by significant intracontinental deformation. A reconstruction using the intra-C34 rotation also illustrates the earliest possibility for direct deep-water connection between the central and south Atlantic oceans. One further consequence of this model is that a total reconstruction derived from it and closure of the central Atlantic between North America and Africa suggests that the Venezuelan and Gulf of Mexico basins may be conjugates formed during the earliest opening of the central Atlantic, and not separate marginal basins.