The Rodinia reconstruction of the Neoproterozoic Supercontinent has dominated discussion of the late Precambrian Earth for the past decade and originated from correlation of sedimentary successions between western North America and eastern Australia. Subsequent developments have sited other blocks according to a distribution of ∼1100 Ma orogenic belts with break-up involving a putative breakout of Laurentia and rapid reassembly of continent crust to produce Gondwana by early Phanerozoic times. The Rodinia reconstruction poses several serious difficulties, including: (a) absence of palaeomagnetic correlation after ∼730 Ma which requires early fragmentation of continental crust although geological evidence for this event is concentrated more than 150 Ma later near the Cambrian boundary, and (b) the familiar reconstruction of Gondwana is only achieved by exceptional continental motions largely unsupported by evidence for ocean consumption. Since the geological evidence used to derive Rodinia is non-unique, palaeomagnetic data must be used to evaluate its geometrical predictions. Data for the interval ∼1150-500 Ma are used here to test the Rodinia model and compare it with an alternative model yielding a symmetrical crescent-shaped analogue of Pangaea (Palaeopangaea). Rodinia critically fails the test by requiring Antarctica to occupy the location of a quasi-integral Africa, whilst Australia and South America were much closer to their Gondwana configurations around Africa than implied by Rodinia. Palaeopangaea appears to satisfy palaeomagnetic constraints whilst surmounting geological difficulties posed by Rodinia. The relative motions needed to produce Gondwana are then relatively small, achieved largely by sinistral transpression, and consistent with features of Pan-African orogenesis; continental dispersal did not occur until the Neoproterozoic-Cambrian boundary. Analogies between Palaeopangaea and (Neo)pangaea imply that supercontinents are not chaotic agglomerations of continental crust but form by episodic coupling of upper and lower mantle convection leading to conformity with the geoid.