Many fault-location algorithms rely on a simulation model incorporating network parameters which closely represent the real network. Estimations of the line parameters are usually based on limited geometrical information which do not reflect the complexity of a real network. In practice, obtaining an accurate model of the network is difficult without comprehensive field measurements of each constituent part of the network in question. Layer-peeling algorithms offer a solution to this problem by providing a fast “mapping” of the network based only on the response of a probing impulse. Starting with the classical “Schur” layer-peeling algorithm, this paper develops a new approach to map the reflection coefficients of an electrical network, then use this information post-fault to determine accurately and robustly the location of either permanent or incipient faults on overhead networks. The robustness of the method is derived from the similarity between the post-fault energy reaching the observation point and the predicted energy, which is based on real network observations rather than a simulation model. The method is shown to perform well for different noise levels and fault inception angles on the IEEE 13-bus network, indicating that the method is well suited to radial distribution networks.