Abstract: There may be too many tuples in an answer after fuzzy query based on relational database. So, it is necessary to rank query results in accordance to user's needs. The ranking process is as follows. According to the degree of membership between tuple and fuzzy query, the tuples with different degree of membership in query results are differentiated from each other to form several membership classes. Then, the improved PIR (probability information retrieval) model and database workload are used to analyze the degree of correlation between the attribute values specified and unspecified in queries, thus ranking the tuples with the same degree of membership due to users' preferences. As a result, the DPR (degree of membership and probability ranking) process is proposed as an automated ranking approach for fuzzy query results, which has been proved to improve greatly the ranking quality of fuzzy query results through experiments and analysis.