Abstract: 3D numerical modeling of tunnel DC resistivity for in-advance detection is carried out in this paper. A fast algebraic multi-grid solver is used to solve the finite difference problem of the second potential. Comparing the numerical solution to the analytical solution of a board model in homogeneous space, the relative error is under 0.28%. Modeling results show that the tunnel cavity itself has influence on the survey data, which could be corrected by ratio curve method. The relative error between the corrected apparent resistivity and the analytical one is 2.4%. Furthermore, a method for identification of the frond anomalies nfluenced by side anomalies is also studied. Modeling results show that the frond anomalies could be correctly identified by surveys on different surfaces of the tunnel or by surveys during the excavation.