The Subaru Deep Field (SDF) project is an ultra deep survey which aims to explore the universe as deep as possible, and as early as possible with Subaru in both visible and near-infrared. In the SDF project, we attempted to take the deepest imaging data among those having a single Suprime-Cam's FOV. The survey goes as deep as the HDF, while it can cover far wider area than the HDF. We believe that this SDF project will provide our community the pride to say that "Subaru could expand our knowledge to such a deep universe." The project guarantees both quality and quantity of the data as well as a plenty of fascinate scientific goals. Although the observational approach is conservative that widespread popularity in current studies of distant universe, the proposed survey based on a large amount of the observational time of an 8m telescope would be a milestone of these deep surveys. It is not too much to say that this project is the "Power Play" of Subaru and the project should be carried out as the observatory key project. The major goal of this survey project is to search and study for the highest redshift (z>6) galaxy population. In this project, we will carry out a couple of different approaches to search them. One is to detect continuum dropout galaxies using deep multi broad-band imaging and the other is the deep narrow-band imaging for a high-z population as having only the Lyα emission feature. This is the first systematic survey to carry on the combination of these two attempts. In addition, this proposed survey will produce our own large sample for galaxies at z=4-6 as well as the nearby (z<1.5) galaxy sample down to M ^*+3. Whole the sample and following careful reduction enable us to quantify the galaxy evolution from z=0 to z=5 with derived fundamental tools such as luminosity function and correlation function. For these scientific goals, we propose as the first step of our survey project to take a deep imaging with Suprime-Cam on B, V, R_C, i', z' bands as well as two narrow-bands. The subsequent step is the following-up spectroscopy observation with FOCAS for detected dropouts, Lyα emitters and so on. The NIR imaging and follow-up spectroscopy would definitely make the field more valuable. This optical imaging data of a single Suprime-Cam' FOV (34' x 27') will be the deepest of all the previous/forthcoming data taken with Suprime-Cam which is an unique instrument of Subaru.