Steganography is the science and art of covertcommunication, which aims to hide the secret messages intoa cover medium while achieving the least possible statisticaldetectability. To this end, the framework of minimal distor-tion embedding is widely adopted in the development of thesteganographic system, in which a well designed distortionfunction is of vital importance. In this paper, a class of newdistortion functions known as uniform embedding distortionfunction (UED) is presented for both side-informed and nonside-informed secure JPEG steganography. By incorporatingthe syndrome trellis coding, the best codeword with minimaldistortion for a given message is determined with UED, which,instead of random modification, tries to spread the embeddingmodification uniformly to quantized discrete cosine transform(DCT) coefficients of all possible magnitudes. In this way, lessstatistical detectability is achieved, owing to the reduction of theaverage changes of the first- and second-order statistics for DCTcoefficients as a whole. The effectiveness of the proposed schemeis verified with evidence obtained from exhaustive experimentsusing popular steganalyzers with various feature sets on theBOSSbase database. Compared with prior arts, the proposedscheme gains favorable performance in terms of secure embed-ding capacity against steganalysis. Uniform Embedding for Efficient JPEG Steganography