In this paper, the diagnosis of soft faults in lossy electric transmission lines is studied through the inverse scattering approach. The considered soft faults are modeled as continuous spatial variations of distributed characteristic parameters of transmission lines. The diagnosis of such faults from reflection and transmission coefficients measured at the ends of a line can be formulated as an inverse problem. The relationship between this inverse problem and the inverse scattering theory has been studied by Jaulent in 1982 through transformations from the telegrapher's equations of transmission lines to Zakharov-Shabat equations. The present paper clarifies and completes the computation of the theoretic scattering data required by the inverse scattering transform from the practically measured engineering scattering data. The inverse scattering method is then applied to numerically simulated lossy transmission lines to confirm the feasibility of the studied approach to soft fault diagnosis.