In order to precisely predict the remaining shelf life of Penaeus Vannamei, the relationship between quality indexes and remaining shelf life and the quality change process of it during the storage process were studied. The sensory and physical-chemical indexes, and microorganisms of P. Vannamei at 277 K, 272.2 K and 255 K were first tested in this study. Then, the prediction models of the shelf life of P. vannamei were established for the comprehensive and some key indexes of its quality, and this were based on both the support vector machine and the BP neural network models. The results showed that the prediction accuracies of the shelf-life prediction models based on the comprehensive indexes of P. Vannamei (97.71% for SVM model and 91.41% for BP model) were higher than those of the prediction models based on several key indexes (84.08% for SVM model and 83.76% for BP model). Meanwhile, the prediction accuracies of the prediction models based on support vector machine (84.08% for key indexes and 97.71% for comprehensive indexes) were higher than those of BP prediction models (83.76% for key indexes and 91.41% for comprehensive indexes). Moreover, the prediction accuracy of the support vector machine (SVM) model based on the comprehensive indexes was 97.71%, which were the highest among the four models. The conclusion also provided a reference for the application of support vector machine and selection of prediction indexes in the shelf-life of other food fields.