The interaction between DNA and proteins comprises a pivotal role in almost every cellular process, including gene regulation and DNA replication. Given a novel protein, it is very important to know whether it is a DNA-binding protein or not and where the binding sites are. Over the last three decades, since the discovery that lac operon was regulated by a protein, knowledge of the DNA-protein interactions has soared. However, it is very difficult to use experimental techniques to identify the DNA-binding proteins because these experiments can be prohibitively laborintensive in studying all the possible mutations of the residues on the molecular surface. Hence, it has been generally recognized that the ability to automatically identify the DNA binding proteins and their binding sites can significantly speed up our understanding of cellular activities and contribute to advances in drug discovery. The main goal of present paper is to review the recent progress in the development of computational approaches to predict DNA-protein bindings. We will show a historical roadmap of the amelioration, and how the modifications promote better performance.