Most of the efforts concerninggraphical representations ofhumans (Virtual Humans) have beenfocused on synthesizing geometryfor static or animated shapes. Thenext step is to consider a human bodynot only as a 3D shape, but as anactive semantic entity with features,functionalities, interaction skills,etc. We are currently working on anontology-based approach to makeVirtual Humans more active andunderstandable both for humans andmachines. The ontology for VirtualHumans we are defining will providethe “semantic layer” required toreconstruct, stock, retrieve, reuse andshare content and knowledge relatedto Virtual Humans.