This article deals with multilingual document indexing. We propose an indexing method based on several stages. First of all the most important terms of the document are extracted using general characteristics of languages and statistical methods. Thus, term extraction stages can be applied to any document whatever the document language is. Secondly, our indexing method uses a multilingual ontology in order to find the most relevant concept representing the document content. Our method can be applied to a multilingual corpus containing document written in different languages. This indexing procedure is part of a Multilingual Document System untitled SyDoM, which manages XML documents.