Recognition and visual discrimination of individual animals permit studies on life history, behaviour and population dynamics without physical capture. Photo-identification is widely used for marine mammals, but its application to dugongs has been limited because identifiable features are rarely visible from boats. Individual identity has instead relied on artificial marks or gene-tags which require direct sampling. Drone technology now allows views of tail flukes and other features from above, creating new opportunities for individual identification. We evaluated the feasibility of using fluke-based photo-ID using a multi-decade photographic database of 764 dugong captures from one population. Variation and persistence in tail fluke morphology were examined in recaptured dugongs with validation from genetic and/or turtle-tag identifiers and high-quality photographs. Fluke-based photo-ID was highly reliable (100% precision, 95.9% recall, 99.1% overall accuracy), with most fluke features remaining stable across multiple years. We then tested drone imagery by blind-matching a subset of flukes (n=25). Drone photographs of sufficient quality enabled correct matching for 79% of dugongs, with mismatches restricted to individuals with low fluke distinctiveness. Fluke-based photo-ID offers a robust, non-invasive method for recognising individual dugongs, and drone imagery extends this approach beyond capture-based surveys when image quality and fluke distinctiveness meet defined thresholds.