Combinatorial auctions , i . e. , auction where bidders can bid on combination of items , is a very impor2 tant application area in the electronic commerce nowadays. It tends to lead to more efficient allocations than tradi2 tional auctions in multi-item auctions , while keeping risks for bidders low. However , the winner determination prob2 lem in combinatorial auctions is NP2hard. By the description of the problem and analysis of its characteristics , this paper proposes a Fitting2First heuristic embedded chaotic search algorithm. Comparing to the traditional algorithms , it is easy to operate and can get better results. The outcome indicates the efficiency and the wide application promise of the heuristic algorithm for solving this problem.