The invention discloses a performance prediction and selection method for a rubber shock absorber based on a BP neural network. The performance prediction and selection method comprises the steps thatthe resonant frequency and the shock absorption efficiency of a specimen of the rubber shock absorber are obtained through a vibration test; on the basis of testing data obtained through the vibration test, taking the geometrical characteristics and shore hardness of the rubber shock absorber as input of the BP neural network, taking the resonant frequency or the shock absorption efficiency of the rubber shock absorber as output of the BP neural network, the neural network is trained, and the mapping relationship between physical parameters and the shock absorption performance of the rubber shock absorber is obtained; and through a neural network model, a large quantity of unknown shock absorbers are subjected to shock absorber performance prediction, then physical variables and shock absorption performance data are managed, and on the basis of the shock absorption performance, the physical variables corresponding to the suitable shock absorber are selected to complete selection. Theperformance parameters of the shock absorber are predicted through the physical characteristic variables of the shock absorber, the shock absorber meeting the requirements only needs to be selected from a database according to the design vibration quantity, generalizability is achieved, and the selection prediction efficiency is high.