Appling the method of risk preferences and attribute reduction to large decision tables with many attributes，respectively，a new strategy is proposed that first classifies decision makers (DMs) and then makes decisions． First，the DMs are classifed into risk-aversion，risk-neutral，and risk-appetite types，then different methods are used to find out the useful criteria corresponding to the different types of DMs，only the useful criteria are used to make decisions． Second，three new risk preference assumptions are established according to the risk preferences of the decision makers． Third，a new method based on advantageous discernibility matrix is proposed to obtain attribute weights． Then，a new method，based on weighted combinatorial advantage values (WCAV) for different risk preference decision makers，for information integration and alternative ranking is introduced． Finally，two real examples with numerical and interval value attribute values are presented to demonstrate the new method，respectively．