The Jurassic–Early Cretaceous fragmentation of Gondwana remains poorly constrained because alternative plate-kinematic models differ in their initial fits and in the interpretation of magma-rich, highly extended crust along southern Mozambique. This study integrates offshore geophysical observations with sedimentological, biostratigraphic and palaeobiogeographic evidence to distinguish direct constraints on marine connectivity from interpretations dependent on a particular reconstruction. Marine faunal exchange and sedimentary records indicate that a continuous trans-Gondwana seaway linking the western Tethys and Palaeo-Pacific through the proto-Weddell Sea system was established at approximately 145 Ma (late Tithonian), implying the loss of any remaining terrestrial connection between East and West Gondwana. Dinoflagellate-cyst assemblages and marine deposition in eastern Indian and western Australian basins indicate a first major terrestrial disconnection between Sri Lanka–India–Madagascar and Antarctica–Australia at approximately 134–133 Ma (latest Valanginian). These two age constraints are largely independent of any single plate model. Near magnetic chron M0r (∼121 Ma), some kinematic models retain a narrow Sri Lanka–Antarctica corridor. Major sea-level falls at approximately 121–120 Ma may have temporarily re-established terrestrial connectivity across shallow parts of this corridor without requiring tectonic re-closing. No direct geological evidence presently confirms such a connection, and its inferred final loss during the onset of transtension and asymmetric crustal accretion at approximately 119–120 Ma remains model-dependent. The synthesis provides a testable chronology for Gondwana fragmentation and distinguishes palaeogeographic inferences constrained directly by geological and fossil evidence from those dependent on plate-kinematic assumptions. It has implications for Southern Hemisphere biogeography, marine circulation and the interpretation of Cretaceous vicariant events.