Including ecology in biodiversity data management systems requires classifications of habitat terms that provide standard definitions and indicate their relationships. In addition to data- bases, a wide range of intergovernmental, conservation and fishery organizations require classifica- tions of habitats and ecosystems to enable comparisons between areas and organize information in maps and reports. However, all of the terms used to describe habitats are concepts whose definition is context-dependent. This paper reviews the key concepts and ecological perspectives involved in classifying marine 'habitats' and 'biotopes' (habitat plus its associated species) so as to advise how they may be used in data management systems. Classifications of biotopes provide practical mea- sures of biodiversity at the ecosystem level. As an example the habitat of a benthic invertebrate is very different in spatial scale to that of a parasite, plankton, tuna or whale. Habitats can be geophys- ical and/or biogenic, and may operate at different spatial scales. For example, aggregations of deep- sea coral colonies <1 m in diameter may form km-scale reefs which contain other habitats (e.g. sedi- ments, sponges). An ecosystem can be physiographically defined as a lagoon, seamount, estuary, abyssal plain or entire ocean. Different sampling methods will define different regions, such as satel- lite images of ocean colour, acoustic maps of the seabed, in situ sampling of water or sediment cores and maps derived from analyses of species distributions that may define biogeographic regions. Because they are sampled (and thus defined) by different methods and can operate at different spa- tial scales, separate classifications are recommended for (1) nekton, plankton and benthos and (2) regions (defined to suit political, geographic or management areas), seascapes (defined by topogra- phy or water mass), biotopes and guilds (e.g. based on body size, diet or sampling method). Further- more, it is recommended to record the measurable features used to describe biotopes (e.g. depth, dominant species, substratum) and to avoid imposing a classification hierarchy where the concepts and methods of defining them are different. Indeed, one can let users create the most parsimonious classification for their purposes.