Aiming at the problem of low signal noise ratio (SNR) in detecting magnetic signal of ship, a detection algorithm based on wavelet transform and RBF neural network was proposed. Based on the characteristic of magnetic signal of ship, the signal was decomposed by wavelet transform, and the low frequency components in last level were taken out to filter out the high frequency noise. Then the RBF neural network was trained by sampled data. The signal polluted by noise was processed by the trained RBF neural network. The results of simulation show that the algorithm increases SNR markedly, and enhances the detection ability of weak magnetic signal of ship.