In the atomic scale, the surface energy anisotropy analysis of 38 surface planes of 10 fcc metals Cu, Ag, Au, Ni, Pd, Pt, Rh, Al, Ir and Pb have been simulated by using the elemental variables φ* and nWS and modified analytical embedded-atom method (MAEAM). The results show that the close-packed surface (111) of fcc metals which have the lowest surface energies will grow preferentially, the surface energies for all the other surface planes increase linearly with cosθ(hkl), where cosθ(hkl) are the angles between the surface planes (hkl) and (111), which is consistent with the experimental and the linear-muffin-tin-orbital atomic-sphere approximation (LMTO－ASA) results. A graphical approach which correctly explains the relation of the surface segregation energy and surface energy is employed. We conclude that the surface segregation takes place or not is mainly determined by the rule that an impurity (solute) with lower surface energy will segregate to the surface of the host (solution) with higher surface energy.