The rough set theory is a mathematical tool which deals with uncertain ,imprecise and incomplete data . It is used widely in artificial intelligence, pattern recognition,data mining, intelligent decisionmaking and other application fields. Attribute reduction is one of the core issues of the rough set theory and it has become a hot topic  of the rough set theory research. This paper briefly describes some recent advancement in the study of attribute reduction algorithms both at home and abroad , focusing on generalizing and analysing the mainstream methods of attribute reduction. Finally the paper discusses the future trend of attribute reduction research.