Abstract: A method for detecting a fault, e.g. a short circuit, break or leakage in an electric cable or conductor, whereby two conductor pairs (1,2 and 4,5), one of which comprises the faulty conductor (1), are connected in such a way as to form a double loop (1,5 and 2,4), a voltage pulse being applied to the intact loop (2,4) to locate the fault (X). Depending on the nature of the defect, the pulse voltage is adjusted to a level sufficient to produce a malfunction, e.g. sparking, in the defective loop (1,5), whereupon the ratio of the currents produced at the ends of this loop is measured and the proportion (U.sub.1 /U.sub.2) is translated into a distance of the fault from the detector.