Spindle assembly is one of the most important components of digital manufacturing equipments. It is key problem to monitor and evaluate its performance to assure the normal process. Aiming at the bearings which damage most easily in spindle system, a new time-frequency analysis method called S transform is studied, time-frequency distribution of vibration signals collected from spindle is obtained, and a singular value decomposition method is employed to condense the time-frequency matrix data so that the fault features can be extracted quantitatively. Simulation and experimental studies have demonstrated that the proposed method may identify the running state of spindle bearing accurately. Thus a new technique for the evaluation of spindle serving performance of digital manufacturing equipments is provided.