With the rapid development and utilization of urban underground space, the phenomenon of excavation adjacent to tunnels is becoming more and more common. Therefore, the influence of excavation on adjacent tunnels has become a serious problem that cannot be ignored in engineering. This study proposed a back propagation neural network (BPNN) to predict the tunnel deformation caused by adjacent excavation. Firstly, the finite element software Plaxis (3D) based on a small-strain stiffness constitutive model was used to generate a complete database of tunnel deformation. The accuracy of the BPNN model was trained and tested by this database. Furthermore, the relative importance of each input variable (i.e., the relative horizontal distance of the tunnel as opposed to a supporting structure, the tunnel buried depth, the maximal displacement of retaining structure) was estimated by this BPNN model. Finally, the prediction of BPNN model showed great agreement with the measured deformations of 17 case histories. The analysis results show that the BPNN model can accurately predict the tunnel deformation caused by adjacent excavation, and can provide a more convenient prediction method for practical engineering.