The fault-generated traveling waves will bounce back and forth between the beginning point and the discontinuity points on the distribution line. The energy of frequency components in the corresponding characteristic which are related to such reflected paths will be significantly higher than that of other frequency components. The traveling waves are decomposed into different frequency bands according to the characteristic frequencies in discontinuity points of the distribution line which includes the branch points, the load terminal points, and the connection points of overhead lines and underground cables. The fault scenarios can be divided into three categories: the fault section around the beginning point, fault section between one and another branch point of the main power wire, and the fault section between one branch point and the terminal load. The energy spectrums within the divided frequency bands vary significantly in different scenarios, and thus these energy spectrum differences can be used to identify the location of fault sections. Once the fault occurred, the similarity between the energy spectrum of the recorded fault-generated waveforms and that corresponds to the typical positions was analyzed. The fault section can then be identified based on the constructed fault location criteria. Since the proposed method is not affected by of the following parameters of the distribution line, including the actual length error, incomplete transposition, and heterogeneous surface, the proposed method provides high reliability in the fault location.The simulation results have proved the accuracy and feasibility of the proposed fault section location method.