Understanding and predicting the distribution of Antarctic toothfish (Dissostichus mawsoni) habitat suitability at different spatial and temporal scales can make an important contribution to CCAMLR’s management of this species. In this paper the use of Maxent models with fishery catch data and environmental data to provide a circumpolar prediction of Antarctic toothfish habitat suitability is explored. Maxent is highly sensitive to biased sampling effort and the selection of background data. Hence, the influence of background data selection on Maxent model predictions using two approaches was tested: (1) a random background sampling approach across the CAMLR Convention Area and; (2) a target-group background sampling approach that aims to correct biased sampling effort. The predictions from the target-group background approach resulted in the most accurate predictions, but predictions needed to be constrained within an appropriate bathymetric range via post-processing. This resulted in an accurate and more realistic prediction of the distribution of Antarctic toothfish habitat suitability.