Objective To investigate the safety of rivaroxaban by using prothrombin time (PT). Methods PT was detected by IL ACL TOP 700 blood coagulation instrument and its supporting reagents. Data were expressed as M (P25~P75). Mann-Whitney U test was used for data comparison between the two groups. Kruskal-Wallis H test was used for age data comparison between multiple groups. P<0.05 was considered statistically significant. The diagnostic performance of PT was evaluated by a receiver operating characteristic curve (ROC). Results In the bleeding group, the PT changes significantly between before and after treatment groups. On the 1st and 3rd days after treatment, the peak PT was significantly higher than the day"s trough level; and on the 3rd day, the peak value and the trough value were significantly higher than the 1st day’s level. There was no significant difference in peak PT between the patients in the non-bleeding group and the bleeding group on the 1st day (P>0.05), while the trough PT in the bleeding group was significantly higher than that in the non-bleeding group (P<0.001). On the 3rd day, the bleeding group the peak and trough PT levels were significantly higher than those without bleeding (P < 0.001). The diagnostic performance of the trough PT predicting bleeding risk on days 1 and 3 after treatment was superior to the peak PT, while the effect of the trough PT predicted on the 1st and 3rd days was similar. In terms of assessing the risk of bleeding, the trough PT is superior to the peak PT, and the 3rd day is better than the 1st day. Conclusion PT can effectively evaluate the bleeding risk of rivaroxaban after sufficient performance verification, and provide a reliable basis for clinical rational drug use.