Due to the excessive width of the main lobe after windowed Fourier transform of the cable input impedance spectrum, there is an excessive ‘dead zone’ in defect positioning near the cable end. To address this issue, an improved method for distribution cable defect positioning based on Chebyshev window is proposed in this paper. Firstly, a distributed parameter model of the cable is established, and the principle of local defect positioning is elaborated based on the cable input impedance spectrum. Subsequently, simulation studies are conducted on the cable input impedance spectrum using Hamming window, Blackman window, and Chebyshev window, comparing and analyzing the effectiveness of each window function. Finally, a diagnostic function is constructed from the cable impedance phase spectrum and combined with the Chebyshev window to achieve defect positioning and defect type identification. Simulation study on a 10 kV XLPE cable, 100 m in length, demonstrates that compared with the direct treatment of cable impedance spectrum with Chebyshev window, the proposed method not only significantly improves defect positioning accuracy: reducing the relative error of defect positioning from 0.467% to 0.029% at 2 m from the cable’s first end, but also shrinking the positioning ‘dead zone’ from 6 m to 3 m away from the cable’s end. Moreover, the proposed method can identify different types of defects according to the waveform characteristics.