Abstract: To enhance the reliability of neutronics simulations for space reactors, this study utilized the Monte Carlo code to model the core configuration of the HMF005 critical benchmark experiment (from the International Handbook of Criticality Safety Benchmark Experiments) and evaluated six core configurations using different nuclear data libraries. The deviations between calculated k_eff values and experimental results were systematically analyzed with statistical evaluations of mean ratios, standard deviations, χ², and root mean square errors. The results indicate that four evaluated nuclear data libraries-ENDF/B-VII.1, CENDL-3.2, ENDF/B-VIII.0, and JEFF-3.3-yield k_eff values within three times the experimental uncertainty. Among these, ENDF/B-VII.1 showed the closest agreement with experimental data, while CENDL-3.2 exhibited slightly larger deviations. Further analysis revealed that the discrepancies in CENDL-3.2 primarily stem from molybdenum (Mo) and beryllium (Be) isotopes, with ⁹²Mo contributing most significantly. Sensitivity analysis demonstrated that the total reaction sensitivity coefficients for ⁹²Mo and ⁹Be are predominantly governed by elastic scattering sensitivities. Notably, a 34.9% discrepancy was observed between the radiative capture sensitivity coefficients of ⁹Be in CENDL-3.2 and ENDF/B-VII.1.