Multiple initiatives have called for large-scale representative networks of marine protected areas (MPAs). MPAs should be ecologically representative to be effective, but in large, remote regions this can be difficult to quantify and assess. We present a novel bioregionalization for the Southern Ocean, which uses the modelled circumpolar habitat importance of 17 marine bird and mammal species. The habitat-use of these predators indicates biodiversity patterns that require representation in Southern Ocean conservation and management planning. In the predator habitat importance predictions, we identified 17 statistical clusters, falling into four larger groups. We characterized and contrasted these clusters based on their predator, prey and oceanographic characteristics. Under the existing Southern Ocean MPA network, some clusters fall short of 10 % representation, yet others meet or exceed these targets. Implementation of currently proposed MPAs can in some cases contribute to meeting even 30 % spatial coverage conservation targets. However, the effectiveness of mixed-use versus no-take MPAs should be taken into consideration, since some clusters are not adequately represented by no-take MPAs. These results, combined with previous studies in the Southern Ocean, can help inform the continued design, implementation, and evaluation of a representative system of MPAs for Southern Ocean conservation and management.