The invention discloses a method for evaluating dangerous chemical safety management. Data mining efficiency is greatly improved, and workload is reduced. Reduction of large scale data is conducted by using a rough set, and the problem of great attribute numbers and excessive large candidate set can be solved. The Apriori algorithm is used to improve operation efficiency, and association rules can be obtained rapidly. Main problems that restrict different enterprises from improving the safety management level are found out, thereby providing theoretical foundation for stipulating daily law enforcement inspection contents for government departments.