The electric spindle is an essential component of modern CNC machine tools as it plays a crucial role in determining the performance and accuracy of the machine. However, the vibration of its bearing unit can significantly impact the spindle’s operational reliability and service life. Therefore, it is essential to develop a reliable method to distinguish between synchronous and asynchronous vibrations at both ends of the electric spindle. In this paper, we propose a signal similarity measurement method based on probability and l2-norm to address this issue. The proposed method employs EMD to decompose the vibration signals of the electric spindle bearings at both ends. Then, probability distribution and l2-norm are used to measure signal similarity effectively. This approach can effectively distinguish between synchronous and asynchronous vibrations at both ends of the electric spindle. Experimental testing was conducted to verify the effectiveness of this method. The results show that the proposed method is reliable and effective in distinguishing between synchronous and asynchronous vibrations at both ends of the electric spindle. This method has the potential to enhance the bearing’s service life and the operational reliability of the electric spindle.