RF Stealth is the most important feature of the 5th generation fighter aircrafts, but their Radar Cross Section (RCS) analyzation is not readily available nor understood by general public. In this paper we analyze the S-band RCS of an F-22 CAD model with Multilevel Fast Multipole Method (MLFMM) and Large Element Physical Optics (LE/PO), and present its monostatic and bistatic RCS, near fields, and surface currents. Although the CAD model is not the real Digital Mock-Up (DMU) of the aircraft, we can still grasp the range and trend of its RCS due to the principle of RCS number's game, and estimate on how high the performances of Radar Absorption Material (RAM) should be in order to meet RF Stealth requirements. Special parts like engine intake and radome are treated as metal in this analysis. We found the monostatic RCS is not as small as generally believed. When RAMs are not applied, the boresight X-Y cut monostatic RCS is around 10 dBsm, and it quickly drops to region between 0 to -10 dBsm between 10° - 40°, and increases to 30dBsm at 90°. The boresight bistatic RCS is at the same value, but generally smaller at other directions. In both monostatic and bistatic RCS, when radar illuminates from boresight direction, the largest RCS appears at the perpendicular and tail direction. MLFMM produces better results than LE/PO. With RAM applied, we believe the RCS can be reduced to the level of -10 ~ -20 dBsm in desired direction and frequency band.