Magnetostrictive transducers have high core eddy current loss, uneven magnetic field distribution and low electromagnetic conversion efficiency under high frequency excitation. These issues need to be addressed by the optimal design of the transducer body. The coil height and yoke loop structure of the transducer are firstly simulated to initially determine the magnetic circuit structure. Then, an overall multi-objective optimization design model for the transducer is proposed, which is based on the non-dominated ranking genetic algorithm. The optimization objectives are to increase the magnetic field strength in the magnetostrictive bar, improve the uniformity of the magnetic field distribution in the bar, and reduce the high frequency loss of the transducer. The normalized ranking and entropy weighting methods are introduced for decision support of the Pareto front solutions obtained by this optimization method to screen a set of optimal design solutions. Finally, the optimal solution is simulated and analyzed. Results of magnetic field distribution and numerical calculation verify the effectiveness of the optimization method.