The invention provides a steganalysis method for detecting a video based on DCT coefficient steganography. According to the method, the influence of a steganographic operation based on the DCT coefficient steganography to space domain and time domain correlation of the video is analyzed, and a space domain feature set is calculated by using a DCT core and an embedded cost to obtain a space domain histogram. Moreover, a space domain fragment is constructed through similar macro blocks between motion vector connecting frames, and a time domain feature set is calculated to obtain a time domain histogram. The space domain histogram and the time domain histogram are merged into a final steganalysis feature set, and the final steganalysis feature set is input to a classifier to be trained to generate a steganalysis classifier. Then, a steganalysis feature of a to-be-tested video is extracted according to the above method, and the steganalysis feature is input to the steganalysis classifier to be analyzed and discriminated. By adoption of the method, steganographic videos based on the DCT coefficient steganography can be effectively detected, and the steganalysis accuracy and particularly the detection accuracy of high-quality videos are improved.