The Norfolk Basin is a small back-arc basin in the SW Pacific with an unknown age and origin for its formation. Moreover, the process behind its formation as a back-arc basin in a region of extensive volcanism has previously never been resolved. We have conducted a detailed interpretation of recently acquired swath bathymetry, gravity and magnetic data from the northern Norfolk Basin (FAUST-2 survey), together with existing data from the region, to address the age and origin of the Norfolk Basin and propose a new model for basin evolution. The Norfolk Basin is characterized by complex bathymetry, which formed during three distinct tectonic phases: (1) regions of elevated topography (termed plateaus) show characteristics of thickened oceanic crust formed during the Cretaceous Normal Superchron, in either a back-arc basin environment or as a trapped piece of Pacific-Phoenix oceanic crust; (2) regions of deep bathymetry (termed basins) separate the Cretaceous plateaus and show basement textured by seafloor spreading fabric of Miocene age; (3) widespread intraplate, plume-related volcanic activity post-dating the formation of the plateaus but predating the formation of the basins. A Miocene model for the formation of the basin regions in the Norfolk Basin requires a 700 km landward jump in the location of back-arc spreading, even with continued subduction at the Tonga-Kermadec Trench and the seaward migration of the subduction hinge. The cessation of spreading in the basin regions would have been followed by an eastward jump of back-arc spreading by about 1000 km, into the current Lau Basin. We propose that the driving force for such an enormous landward jump in back-arc spreading is active rifting driven by a mantle plume. This scenario is supported by the extensive Miocene intraplate volcanism observed in the Norfolk Basin and the elevated bathymetry of the entire region. In this model, mantle driven uplift and excess volcanism would have favored back-arc rifting landward, rather than seaward, of the existing South Fiji Basin, in an extensional regime on the over-riding plate caused by the continued roll-back of the subduction hinge. This scenario is supported by the numerous Miocene volcanic cones in the Norfolk Basin, which are mainly located on the preexisting plateaus, probably predating the formation of the basins, as well as their plume melt signature. The Norfolk Basin appears to be the first known example of a major back-arc basin whose position relative to the slab was not controlled by subduction-driven mantle flow and melting but by plume tectonics. This requires a re-evaluation of the traditional back-arc basin theory in supra-subduction zones.