In order to accurately grasp the spatial distribution of soil cadmium in the Yellow River Basin, different combinations of environmental factors and soil physicochemical factors were used as auxiliary variables, and the genetic algorithm (GA) was used to optimize the radial basis function (RBF) neural network to predict the spatial distribution of soil cadmium in the Yellow River Basin, and the prediction accuracy of this model was compared with those of the regression Kriging and the RBF neural network, to investigate the effects of soil physicochemical factors and GA on the prediction accuracy of the neural network. The results showed that: 1) The addition of soil physicochemical factors (organic matter content, pH, CEC) could improve the prediction accuracy of the neural network model. The root mean square error (RMSE), mean absolute error (MAE), and mean relative error (MRE) of the GARBF neural network model based on the environmental factors and soil physicochemical factors were reduced by 0.058 mg/kg, 0.033 mg/kg, and 4.4 percentage points, respectively; 2) GA could improve the prediction accuracy of neural network models, and the RMSE, MAE, and MRE of the GARBF neural network model based on environmental factors and soil physicochemical factors were reduced by 0.009 mg/kg, 0.005 mg/kg, and 0.6 percentage points, respectively, compared with the RBF neural network model based on environmental factors and soil physicochemical factors. 3) The prediction results obtained by adding environmental factors and soil physicochemical factors and optimizing the neural network model using GA were optimal, and the GARBF neural network model based on environmental factors and soil physicochemical factors could be used to predict the spatial distribution of soil cadmium in the Yellow River Basin.