Abstract: Aiming at the randomness and imperceptibility of chatter marks in a 2180 mm tandem cold rolling mill, two methods to identify the vibration of the rolling mill were systematically studied based on wavelet analysis and the fractal theory. Vibration signals from the rolling mill under typical conditions were analyzed by the two methods. The results show that both can effectively identify the vibration of the rolling mill. This paper provides an effective way to identify the chatter vibration in the complex vibration environment. The two methods have practical significance for real-timely monitoring the operation condition of rolling mills, avoiding the occurrence of hazardous production accidents, and then predicting the vibration of rolling mills.