To explore impact on marine habitats caused by coastal pollution, predictive models of the Indo-Pacific humpback dolphin (Sousa chinensis) distribution were generated using the Maxent modelling approach, based on dolphin occurrence data in Sanniang Bay and its adjacent waters in Guangxi Province during 2006-2007, the independent oceanographic parameters derived from Google Earth Digital Elevation Model (DEM) and Landsat images, and the water environmental data from the marine investigation in 2012. Results indicated that the distance to the major river mouths was the strongest predictors, and water nutrition and persistent pollutants were the important factors for the dolphin distribution in both wet and dry seasons. Based on the results, habitat selection strategies of the dolphin were discussed: the dolphin tended to select estuaries with abundant prey and avoid highly polluted areas, which could be useful in designing Marine Protected Areas (MPAs) and formulating management strategies.