Methods and apparatus, including computer program products, implementing and using techniques for generating a file plan and a thesaurus for a records management system. A first user input defining one or more top level terms to be used in the file plan is received. A second user input defining one or more child level terms to be used in the file plan for at least some of the top level terms is received. A third user input defining one or more term relationships among the collection of child level terms and the top level terms is received. Without further user input, a file plan and a thesaurus are generated based on the top level terms, the child level terms and the specified term relationships among the child level terms and top level terms.