Efforts are currently being pursued to validate present-day analysis methods for light water reactors (LWRs) in conjunction with advanced fuels having a high plutonium content. One particular problem, encountered in the framework of cell calculations performed at the Paul Scherrer Institute for infinite arrays of uranium-free plutonium fuel rods considered earlier in the framework of an international benchmark, is that significant *k*_{} decreases can be observed by just raising the upper energy boundary of the thermal range from 1.3 to 2.4 eV (some other LWR lattice codes utilize even smaller values for this cutoff energy). The present study indicates that the sensitivity of the computed *k*_{} with respect to the upper boundary of the thermal range results primarily from the different scattering matrices for hydrogen. The thermal motion of the scattering nucleus is taken into account in the thermal energy range, while a zero velocity of the scattering nucleus is assumed in the elastic scattering matrices employed in the epithermal energy range. Therefore, if the thermal cutoff energy is increased, the energy loss of the neutrons scattered down by hydrogen from energies between the original and the larger cutoff value is reduced, resulting in a shift of the neutron spectrum toward higher energies. The sensitivity of the computed *k*_{} with respect to the upper boundary of the thermal range, however, does not depend on upscattering effects, which were estimated to be negligibly small for energies greater than ~1 eV.

Instead of further raising the upper energy boundary of the thermal range (the choice of an appropriate, sufficiently large value being strongly problem dependent), the analytical recalculation of the epithermal scattering matrices for the moderator nuclides, based on the free gas model (instead of the elastic scattering model, which assumes a zero velocity for the scattering nucleus), is proposed. Several beginning-of-life LWR systems have been analyzed with the new method. In particular, for a representative cell with uranium-free plutonium fuel consisting of a mixture of oxides of reactor-grade plutonium, zirconium, and the burnable poison erbium, the use of the new epithermal free gas scattering matrices results in a *k*_{} decrease of ~700 pcm (pcm = 1.0 × 10^{-5}). For a mixed-oxide (MOX) fuel assembly with 4.8 wt% fissile plutonium in depleted uranium, typical of a present-day pressurized water reactor, the computed *k*_{} decreases by as much as ~400 pcm. For fresh assemblies with UO_{2} fuel, the decrease is less significant (<100 pcm). The effect is thus particularly important for cases with strong resonance absorptions in the lower-electron-volt range. Bearing in mind that present-day LWRs are being loaded with a larger number of assemblies with MOX fuel, a general recommendation is to compute the epithermal scattering matrices of hydrogen by accounting for the thermal motion of the nucleus below a sufficiently high energy limit. Upscattering effects being negligibly small for energies greater than ~1 eV, the specific choice of a thermal cutoff energy larger than ~1 eV is, of course, not important as far as upscattering is concerned. However, the choice of a sufficiently high thermal cutoff energy could help in reducing the inaccuracies produced by an inadequate model in the epithermal energy range.