Single-crystal Fe16N2 films have been grown epitaxially on Fe(001)/InGaAs(001) and InGaAs(001) substrates by molecular beam epitaxy (MBE). Saturation flux density B(s) of Fe16N2 films has been demonstrated to be 2.8-3.0 T at room temperature, which is very close to the value obtained by Kim and Takahashi using polycrystalline evaporated Fe-N films. Temperature dependence of B(s) has been measured. B(s) changed with temperature reversibly up to 400-degrees-C, while beyond 400-degrees-C, B(s) decreased irreversibly. X-ray diffraction showed that Fe16N2 crystal is stable up to 400-degrees-C, while beyond 400-degrees-C, Fe16N2 dissolves into Fe and Fe4N, and also some chemical reactions between Fe16N2 and the substrate occurs. This caused the temperature dependence of B(s) mentioned above. From the temperature dependence of B(s) up to 400-degrees-C, the Curie temperature of Fe16N2 is estimated to be around 540-degrees-C by using the Langevin function. The above mentioned B(s) of 2.9 T at room temperature and 3.2 T at - 268-degrees-C corresponded to an average magnetic moment of 3.2-mu-B per Fe atom and 3.5-mu-B, respectively. These values of the magnetic moment of Fe atoms are literally giant, far beyond the Slater-Pauling curves. The origin of the giant magnetic moment has been discussed based on the calculation carried out by Sakuma. However, there was a significant disagreement between experimental values and calculated ones, so the origin remained to be clarified. Also, magneto-crystalline anisotropy of Fe16N2 films has been investigated.