Aiming at working out the problem of existing load balancing algorithms that failed to consider the requirements of different types of services sufficiently in heterogeneous wireless networks, a new load balance scheduling algorithm was proposed in this paper. Firstly, three utility functions were designed to reflect three implementation-specific aspects, including the effective bandwidth, real-time calls blocking rate and the average non-real-time calls transmission time as per the needs of different types of services combined with the actual performance of each available network in heterogeneous wireless networks. Further, a fuzzy logic algorithm was adopted to fulfill the fuzzy decision of three utility functions above. Finally, we adjusted the load of each network according to the fuzzy decision value to achieve equilibrium. Simulation results showed that the proposed load balancing algorithm performs much better than reference algorithms with respect to the system blocking probability, average transmission time and network load rate equalization.