In combinatorial auctions buyers nay bid for arbitrary combinations of goods. But determining the winners of combinatorial auctions who maximize the profit of a seller is known to be NP-complete. A branch-and-bound method can be one of practical algorithm for winner determination. However, bid selection heuristics play a very important role in the efficiency of a branch-and-bound method. In this paper, we designed and implemented an algorithm which used a branch-and-bound method and Linear Programming for winner determination in combinatorial auctions. We propose new bid selection heuristics which consider a branching bid and conflicting bids simultaneously to select a branching bid in the algorithm. In addition, upper bounds are reused to reduce the running time in specific cases. We evaluated the performance of the algorithm by experiments with five data distributions and compared our method with others. The algorithm using heuristics showed a superior efficiency in two data distributions and a similar efficiency in three distributions.