Abstract: In this paper, an image steganographic algorithm preserving the first-order statistical properties is proposed. We firstly analyze the state transition model of LSB-based algorithms and derive the statistical properties of stego-image, then the message embedding method based on histogram modification is built. It has the good character of preserving first-order statistical properties and can be implemented conveniently. Experimental results show that this approach achieves good performance of avoiding detection, existing steganalysis techniques against LSB-based algorithm such as RS analysis and sample pair analysis failed to detect the existence of hidden messages. This approach is also applicable for color images.