In general, sedimentary basins are characterized by negative free-air and Bouguer gravity anomalies. However, the extensional basins of the Ross Sea are paradoxical in that positive gravity anomalies overlay the Victoria Land Basin, Northern Basin, Central Trough and Northern Central Trough while basement highs are associated with negative gravity anomalies. Measured basement densities from DSDP basement cores give values between 2600–2800 kg/m 3 while bulk sediment densities range from 1210–2200 kg/m 3, indicating a normal density relationship between basement and sediment infill. In contrast, the relatively young and narrow Terror Rift is associated with negative free-air and Bouguer gravity anomalies, but has a different geological history as compared to the larger Ross Sea basins. Process-oriented gravity modeling indicates that magmatic underplating and crustal intrusions are inconsistent with the observed gravity and basement geometry of the Ross Sea basins. The magma volume necessary to account for the distribution and amplitude of the positive gravity anomaly of the Central Basin and be isostatically balanced would need to be comparable to the tholeiitic flood basalt volume of the Columbia River province—it is thus unlikely that the volume of Neogene volcanics of the Ross Sea region is sufficient to explain the observed gravity relationship by modifying the bulk density of the crust. We demonstrate that positive free-air and Bouguer gravity anomalies over extensional basins are the consequence of a relatively low flexural strength of the lithosphere during rifting being contrasted by higher flexural strengths later during sedimentation. As the difference between the rigidity of the lithosphere during sedimentation increases relative to the rigidity of the rifted lithosphere, the gravity over the basin becomes progressively more positive but only for a limited range of wavelengths. The narrow width of the Terror Rift precludes it from having a positive gravity anomaly while the opposite is true for the large Ross Sea basins. For the Ross Sea region, such a loading scenario requires a significant delay between extension and the timing of sediment infilling of the basins, consistent with the late Cretaceous extension of the Ross Sea region and the sedimentary succession being dominated by large-scale late Eocene–Neogene glaciogenic progradational sequences. Sediment source was presumably from the denudation of the Transantarctic Mountains, which commenced in the late Paleogene. The time delay between the late Cretaceous formation of the Transantarctic Mountains, late Paleogene exhumation, and the generation of significant Paleogene paleobathymetry requires either the Ross Sea region to be sub-aerial and sediment starved for most of the Paleogene and/or the Paleogene climate was ineffective in producing clastics until the onset of glaciation in the late Eocene–early Oligocene.