Abstract: In order to quantitatively evaluate the fatigue damage of used parts, standard tensile test specimens were made from the same material as sample parts in the research object of hydraulic cylinder. After fatigue tensile tests on the specimens, the surface metal magnetic memory signals were analyzed at different fatigue stages. The eigenvalues and evaluation model are proposed to characterize the degrees of damage. The results show that after determining the stress concentration location according to the positioning index, the degree of damage is estimated based on the evaluation index. Metal magnetic memory detection technology provides a fundamental method for the quantitative evaluation of re-manufacturing ability of fatigue damage of used parts.