Asymmetric sea-floor spreading has occurred south of Australia within narrow longitudinal zones of the Southeast Indian Ocean. The process of asymmetric spreading can be shown to be continuous at the scale of about 10 km or less in contrast to a process involving large, discrete jumps of the ridge-crest position. As there are no discontinuities in the total opening rates between the Australian and Antarctic plates, discontinuities in one-limb rates result when asymmetric and symmetric spreading occur in adjacent zones, and ridge-crest offsets (transform faults) are formed as a consequence. In this case, the length of the transform zone changes with time, and the ridge—fracture-zone system does not record the geometry of the initial rift. Other large areas of the world's ocean basins, in particular the northeast and southwest Pacific, have probably undergone similar phases of tectonic evolution. By relating minimum crustal strength to maximum ambient temperature, the preferred locus of asymmetric splitting implies a process of effective asymmetric cooling at the accreting boundary. A simplistic asymmetric cooling model is both quantitatively and qualitatively investigated. The indications of steady-state spreading conditions, which persist for long periods of time but which change to new steady-state conditions quickly, and the existence of abrupt geographic spreading provinces are difficult to reconcile with any thermal models. Variations in the spatial and temporal spreading evolution of the Southeast Indian Ocean are qualitatively examined by considering the effects of a hypothetical heat source or sink that moves relative to the accreting margin.