An improved MEW （ muhiplicative exponent weighting） algorithm, SLE-MEW is proposed for vertical handoff decision in heterogeneous wireless networks. It introduces the SINR（ signal to interference plus noise ratio） effects, LS （least square） and information entropy method into the algorithm. An attribute matrix is constructed considering the SINR in the source network and the equivalent SINR in the target network, the required bandwidth, the traffic cost and the available bandwidth of participating access networks. Handoff decision meeting multi-attribute QoS（quality of serv- ice） requirement is made according to the traffic features. The subjective weight relation of decision elements is determined with LS method. The information entropy method is employed to derive the objective weights of the evaluation criteria, and lead to the comprehensive weight. Finally decision is made using MEW algorithm based on the attribute matrix and weight vector. Four 3GPP（ the 3rd generation partnership project） defined traffic classes are considered in performance evaluation. The simulation results have shown that the proposed algorithm can provide satisfactory performance fitting to the characteristics of the traffic.