With the growing popularity of XML technology, developing effective and efficient query techniques of XML data becomes an important problem. In this paper, we introduce a query recommendation technique, which is able to provide both result recommendation and query input recommendation for XML queries, to help users to find useful information more efficiently. In our approach, we mine association rules from XML documents to discover related XML fragments and recommend to the user the possibly interesting results, which are however not included in the original query results. In order to restrict the number of discovered rules and to discard uninteresting rules, our system allows users to specify constraints on the generated association rules. In addition to the recommendation of results, our system also provides potentially relevant XML queries to help users to find the desired content rapidly. Our experimental results show that our methods can provide recommendations for both query results and query input effectively.