This paper discusses the adsorption of metal single-atoms (SAs) on graphitic C3N4 (gCN) multilayered surface models based on density functional theory (DFT) calculations. We systematically searched stable adsorption sites on gCN and showed that the cavity site is preferred for metal SAs. We also discussed the usefulness of several descriptors, such as the distance between the SA and gCN, charge donation (electron transfer), and atomic radius, for evaluating the SA adsorption energy, based on the linear regression method. In addition, the most stable adsorption energy for fifth- and sixth-row metals can be predicted using only the “group” descriptor of the periodic table. The group descriptor is closely related to the atomic radius, and hence the stability of fifth- and sixth-row metals is determined from the fit between the SA radius and the gCN cavity size. The study findings facilitate the development of various metal-dopant techniques for gCN-based applications.