Abstract: In order to improve the accuracy of spindle detection for precision machine tools, a method was proposed based on current for evaluating the performance degradation of spindle. A performance degradation model of the spindle was established to facilitate the monitoring and evaluation of the spindle condition. Firstly, the wavelet packet threshold was used to denoise the current signal, and then the multi domain feature space was constructed by extracting the time-frequency features of the denoised current signal. Then the principal component analysis (PCA) was used for data dimensionality reduction, and the dimensionality reduction of samples was used for support vector machine regression modeling. the particle swarm optimization (PSO) algorithm was used to optimize the parameters of the support vector machines (SVM) model to obtain the optimal performance degradation model. Finally, the model was applied to the evaluation of spindle performance degradation in a experiment platform. The experimental results show that the method is correct and can accurately evaluate spindle performance.