The invention discloses a mountainous area 10kV line-to-ground resistance design method based on a self-adaptation neural network. The method comprises the steps that a soil resistance gauge is adopted for measuring multiple sample data of the layered resistance and the layered soil thickness of a region where a 10kV overhead line structure is located, the sample data is adopted as input signals of the self-adaptation neural network, the equivalent soil resistance is adopted as output signals, a population optimization method is used for optimizing a connection weight among a self-adaptation neural network input layer, a hidden layer and an output layer, after iterative learning, an equivalent soil resistance optimization value is obtained, and then the optimized ground resistance is obtained through a ground resistance design formula. By adopting the method, accurate calculation of the overhead line structure ground resistance under the mountainous complex geologic environment can beachieved, and the accurate design and construction bases are provided for the 10kV overhead line anti-thunder project. The method has the advantages, not achieved in the prior art, of being higher incalculation precision and easier in calculation method project implementation.