Abstract: Users often have fuzzy or imprecise requests when querying XML documents. A new approach based on XML content and structure was proposed to reflect users' fuzzy query intention. Based on the fuzzy set theory, a fuzzy extension of XPath query expression was proposed, which can be expressed exploiting fuzzy predicates. And then fuzzy query relaxations was provided to get more querying results which satisfy users' query requests. The proposed scoring method uses an extended vector space model, which considers the relevance of both content and structure when ranking these query results. According to the matching of the structure and content, the nodes whose scores are greater than the threshold are query results. Finally, the efficiency of the approach is demonstrated by experimental results.