A new technique is proposed to detect and locate wire faults using the impedance spectroscopy and a model-based approach. A modeling approach for transmission lines is developed, in which every cable part having the same characteristic impedance is represented analytically by a frequency-dependent ABCD model. The model parameter extraction is resolved by global optimization techniques based on genetic algorithms leading to a robust convergence behavior and excellent accuracy. This novel method enables the location of hard and soft faults and the identification of types of wire faults. The bandwidth of the developed transmission line model fits to experimental results, so that influence effects such as losses, dispersion, and frequency-dependent signal propagation can be precisely modeled. The calculation time is not proportional to the wiring length or dependent on cable system complexity by numerical methods such as FDTD.