The invention discloses a robust video steganography method for reducing H.264 interframe drift distortions. The method comprises the following steps: step 1, analyzing the interframe drift distortions to obtain the cause of the drift distortions, then classifying the generated drift distortions according to the size, and selecting an embedding position of a DCT brightness coefficient according tothe size of an embedded drift distortion; step 2, carrying out convolutional encoding on an information sequence to be embedded, wherein each input code element corresponds to two output code elements; step 3, embedding an output sequence into the DCT coefficient, after the embedding is completed, performing entropy coding on an H.264/AVC video stream to form a dense video sequence; and step 4, performing entropy decoding on the H.264/AVC video stream, extracting an original DCT coefficient, acquiring the output sequence, decoding the output sequence, outputting the information sequence to beembedded, and completing the extraction of the information sequence to be embedded. The robust video steganography method for reducing the H.264 interframe drift distortions provided by the inventionimproves the data embedding capacity, the video quality and the robustness.