During an extended LOCA in a LWR, the current UO2 fuel reaches very high temperatures and eventually melts, while the current Zircaloy fuel cladding oxidizes releasing hydrogen. These two consequences can lead to an unacceptable amount of radioactive release by presenting accident routes for containment failure. After such an accident at the Fukushima NPP in 2011, the development of Accident Tolerant Fuels LWRs gained additional momentum which aims to increase the margin to fuel melting, and to preserve cladding integrity as long as possible. Among the top ATF candidates compounds are UN and U3Si2, which have a high thermal conductivity and high uranium density. UN melts at 2850 °C on par with UO2, while U3Si2 melts at only 1665 °C. U3Si2 may serve as a second phase in UN–U3Si2 composites with better material properties than pure UN. Early studies on powders and dense samples, found the chemical UN corrosion by steam at all T,p pairs to generate a sandwiched UN/α‒U2N3+x/UO2 corrosion layer with inferior density. It was seen that dense polycrystalline UN would perform poorly due to an intergranular cracking mechanism due to the stresses caused by the growth of this layer. Due to the missing technological ability to control parameters like grain size and open porosity no work exist on the microstructure dependence of high density UN pellet corrosion in steam, and the intergranular cracking mechanism was never captured by imaging techniques. Also, as UN–U3Si2 composites are fairly new, the degradation mechanism of high density samples under steam is not known as no degradation study has yet been published. This work aimed at increasing understanding of the high pressure steam degradation mechanism of Spark Plasma sintered, microstructure controlled, UN and UN–10U3Si2 (wt %) composite samples, and to analyze the influence of grain size and density on the UN corrosion rate. A further goal was to image corrosion progress in the microstructure. For this, HPBAC steam exposure tests on UN and UN–10U3Si2 at 303 °C and 9 MPa were done. Durations were up to 1.5 hours. Samples were 96–99.9% of theoretical density and grain sizes were 6–24 µm. The corrosion in the microstructure of all samples is imaged by Light Optical Microscopy (LOM). Scanning Electron Microscopy/Energy Dispersive Spectroscopy (SEM/EDS) were used to track the change in chemical composition at the grain boundaries. Two continuous steam exposures in flowing Ar and N2 at 400 °C and 1 atm have been done to study the role of N2 and NH3 on the degradation. One TGA on the residue of one of the autoclave tests was done to confirm the final oxidation state. TGA confirms that at 303 °C and 9 MPa the final product is UO2, while Digerator results show that under N2 the corrosion is faster. LOM and SEM/EDS show that UN–pellets exposed to steam are breaking apart by intergranular cracks generated by a layered precipitation of U2N3+x/UO2 in the grain boundaries. As the density of the products differs greatly from that of UN, high intergranular stresses result in cracking. Cracking makes progressively more surfaces available to oxidation/hydrolysis. An increase in density and reduction of open porosity slows the corrosion process, while an increase in grain size accelerates the degradation. Consequently, all other considerations cast aside the most waterproofed microstructure of a pure polycrystalline UN sample will have maximized density, eliminated open porosity, while maintaining a small grain size. As clusters of UN grains are enveloped by the U3Si2 phase in UN–10%U3Si2, the cracking was seen to be predominantly intragranular. Irrespective of the quality of the microstructure polycrystalline UN will fail by intergranular corrosion. U3Si2 seems to react preferentially with the steam precipitating UO2, delaying the attack on the UN grains. The low degree of maximum weight gain and different corrosion progression in the microstructure of UN–10U3Si2 are strong indications that the composite may provide significantly higher steam tolerance than pure UN.