We propose a novel diagonal co-clustering algorithm built upon the double Kmeans to address the problem of document-word co-clustering. At each iteration, the proposed algorithm seeks for a diagonal block structure of the data by minimizing a criterion based on the variance within and the centroid effect. In addition to be easy-to-interpret and efficient on sparse binary and continuous data, Diagonal Double Kmeans (DDKM) is also faster than other state-of-the art clustering algorithms. We illustrate our contribution using real datasets commonly used in document clustering.