There are lots of ranking algorithms used in Web information retrieval. However, current algorithms have some problems: these algorithms are based on different calculation formulas to calculate the documents and query similarity or train a lot of training data to get corresponding calculation formula which calculate documents and query similarity. We know that this process is a very complex, and sometimes the accuracy is not good. So we propose a new algorithm — dominating ranking algorithm for information retrieval, which rank documents according to the number of dominated documents and the dominated documents' dominating capacity. Experimental results show that the algorithm proposed in this paper could achieve the best performance on the TREC Web Track data set.