Abstract: The rapid deceleration behavior is an important factor affecting safety and efficiency. Based on the vehicle trajectory data obtained by the controller area network (CAN) equipment, this paper designed a threshold method for diagnosing the rapid deceleration behavior. Then, the characteristics of the vehicle rapid deceleration were analyzed from the perspectives of initial speed, deceleration duration, and average deceleration. Further, the correlation characteristics between the rapid deceleration and the traffic state were excavated. Finally, the hot-spot identification of the rapid deceleration and the clustering analysis of rapid deceleration behavior were discussed. Results show that the shorter the duration of the rapid deceleration, the greater the average deceleration; the greater the initial speed, the greater the average deceleration. There is a significant positive correlation between the occurrence probability of the rapid deceleration and the traffic congestion; besides, the spatial location of the rapid deceleration is closely related to road traffic nodes. The K-means cluster analysis shows that the rapid deceleration characteristics of different driver types are significantly different.