Background: Conservation planning and the design of marine protected areas (MPAs) requires spatially explicit information on the distribution of ecological features. Most species of marine mammals range over large areas and across multiple planning regions. The spatial distributions of marine mammals are difficult to predict using habitat modelling at ecological scales because of insufficient understanding of their habitat needs, however, relevant information may be available from surveys conducted to inform mandatory stock assessments. Methodology and Results: We use a 20-year time series of systematic aerial surveys of dugong (Dugong dugong) abundance to create spatially-explicit models of dugong distribution and relative density at the scale of the coastal waters of northeast Australia (~136,000 km²). We interpolated the corrected data at the scale of 2 km * 2 km planning units using geostatistics. Planning units were classified as low, medium, high and very high dugong density on the basis of the relative density of dugongs estimated from the models and a frequency analysis. Torres Strait was identified as the most significant dugong habitat in northeast Australia and the most globally significant habitat known for any member of the Order Sirenia. The models are used by local, State and Federal agencies to inform management decisions related to the Indigenous harvest of dugongs, gill-net fisheries and Australia's National Representative System of Marine Protected Areas. Conclusion/Significance: In this paper we demonstrate that spatially-explicit population models add value to data collected for stock assessments, provide a robust alternative to predictive habitat distribution models, and inform species conservation at multiple scales.