In this study, with the FEM we analyzed the variation of the resonance frequency, bandwidth, and sound pressure of the Tonpilz transducer in relation to its design variables. Through statistical multiple regression analysis of the results, we derived functional forms of the resonance frequency, bandwidth, and sound pressure in terms of the design variables. By applying the constrained optimization technique, SQP-PD, to the derived function, we determined the optimal structure of the transducer that could provide the highest sound pressure level at the resonance frequency of 30,000 Hz and having the -3 dB bandwidth more than 10%, The validity of the optimized results was confirmed through comparison of the optimal performance with that of the FEA. The optimal design method proposed could reflect all the cross-coupled effects of multiple structural variables, and could determine the detailed geometry of the transducer with great efficiency and rapidity.