In this paper, a reversible data hiding scheme based on multiple residual histogram shifting is proposed. As is known to all, there is a prediction technology which can improve the hiding ability under the premise of ensuring good stego-image quality. It studies the similarity between image pixels and uses the residual histogram of the original coverage image to hide secret data. However, when the residual histogram is used to hide the secret data, some pixels in the original image are not well utilized. To solve this problem, we proposed a novel reversible data hiding scheme based on multiple residual histogram shifting. The basic image composed of basic pixels is extracted and processed by prediction technology, and the residual histogram of the basic image is used to hide the secret data too. The experimental results demonstrate that the proposed scheme has higher hiding capacity and better stego-image quality.