Improper operations during cable installation and adverse operating environments often lead to various cable defects. Failure to promptly address these defects can pose significant risks to urban power supply systems. The frequency domain reflectometry (FDR) technique is widely employed for the detection and localization of various minor defects in power cables. However, the current methods face challenges in quantitatively assessing the severity of these defects. This paper proposes an extended Prony method based on a parametric model for the quantitative evaluation of cable defects. The identified defects are first subjected to parameter fitting using the extended Prony method, and the singular value decomposition-least squares method is then used to select the appropriate model order for fitting. Various defects and their severities are simulated, and after normalizing the localization spectrum, the extended Prony method is employed for parameter fitting, leading to quantitative evaluations of defects with different severities. Simulation results for cables with single and multiple defects demonstrate the effectiveness of the proposed extended Prony method in fitting the amplitude at defect sites.