Abstract: A new knowledge-based approach to fuzzy query translation for a database is proposed to overcome the shortcoming of conventional DBMS (database management system) that does not support the fuzzy queries over any of databases. Based on the theory of fuzzy sets, the formation conditions for fuzzy queries over database are described. Then an approach is presented to leverage the membership functions, domain knowledge and α-cut operation of fuzzy numbers for fuzzy query translation. The importance of query criteria and users' preferences are taken into consideration when a fuzzy query is expanded and translated. Neither SQL language nor the database engine requires modification to implement fuzzy queries by the approach proposed, thus enabling users to improve their interaction with the system.