The invention belongs to the electric power system load prediction and power-distribution capacity-to-load ratio configuration technology field and especially relates to a power-distribution optimal capacity-to-load ratio determination method considering an electric vehicle charging demand. The method comprises the following steps of selecting an LS-SVM input vector of an electric vehicle load; establishing a basic LS-SVM space load prediction model; considering the electric vehicle charging demand, and establishing an LS-SVM integrated load forecasting model of an expanded factor; processinghistorical data in the extended LS-SVM integrated load forecasting model, and selecting optimal parameters of the model according to s Bayesian framework; according to a parameter optimization result,determining a model content and carrying out load forecasting on a specified area; evaluating a basic power load of a certain area; considering a predicted load of the electric vehicle, and acquiringa composite peak value of electricity consumption; and establishing a capacity-to-load ratio optimization model of a power distribution network under multiple voltage levels, and determining the optimal capacity-to-load ratio. The charging load of the electric vehicle can be accurately predicted, the power-distribution optimal capacity-to-load ratio is determined in consideration of an electric vehicle charging demand, and a high application value is possessed.