This paper introduces a design method for polyvinylidene fluoride (PVDF) structural acoustic sensors for the active control of sound radiation into enclosures. It combines genetic algorithms and the quadratic optimal approach to search for a sensor configuration capable of detecting vibration components with strong sound-radiation ability. In this research, one PVDF sensor is not limited to one single piece of continuous PVDF film. It can consist of a cluster of small PVDF pieces, which could be discrete. Therefore, the parameters to be optimized are the number and the locations of PVDF pieces involved in a sensor. The design method is applied to a cylindrical shell with a floor partition. The general design guidelines are discussed. To show the effectiveness of the method, the control performance of an optimal sensor arrangement is compared with that of non-optimal ones. Physical insights are obtained using structural modal response analysis, modal spectrum analysis of the PVDF sensor output, and structural acoustical coupling analysis. The performance of a PVDF sensor configuration designed at one acoustic resonant frequency is also investigated for other disturbance frequencies below 500 Hz, showing that a significant reduction of acoustic potential energy can be achieved over a wide frequency range. It is demonstrated that, with PVDF sensors optimally designed using the proposed method, the active control of sound radiation into enclosures can be achieved without using acoustic transducers.