Support vector machine(SVM) is a new learning machine based on the statistical learning theory. For it not only has solid theory, but also can solve many practical problems in optimizing area, such as small samples, over learning, high dimension and local minima. And has shown attractive potential and promising performance in a wide range of fields and applications. In this paper, the least square support vector machine(LS-SVM), which developed from normal SVM, has been used for predicting and modeling the underwater acoustic performance of sandwich composite plates, which were composed of fiber-reinforced plastics/acoustic rubber/fiber-reinforced plastics. Effective result indicate that LS-SVM is of potential application in sandwich structural composite design and acoustic material research.