The invention discloses a method and system for judging traffic jams based on videos. Images are extracted from videos in a timed mode, feature points of each frame of image are extracted, the vehicle speed is calculated by utilizing the displacement difference of the feature points, the queuing length can be calculated according to the feature points, and the traffic jam conditions are judged according to the relations of the vehicle speed, the queuing length and a set threshold. Since the feature points can be extracted under any condition and are not subjected to influence of moving objects and environmental interference, e.g., the number of the feature points and the location of the feature points are not affected in various environments including shadow, rainy days and dusk, the traffic jam judging accuracy of the method is improved greatly. Moreover, the time consumption of programs is controlled by controlling the number of the feature points, and under general conditions, the consumed time can be saved by a half by utilizing the feature point detection compared with a background foreground method. Therefore, the real-time performance and accuracy of the method for judging traffic jams are improved greatly.