To validate new nuclear cross-section evaluations and computational methods, a large number of benchmark experiments were performed in the past. The Organisation for Economic Co-operation and Development (OECD) Nuclear Energy Agency (NEA) launched several projects aiming to collect, preserve, and disseminate the benchmark data in a user-friendly format. Reactor physics benchmarks are covered by the International Reactor Physics Experiments (IRPhE) project. This paper presents the preparation of the IRPhE compilation for the KRITZ-2 critical experiments, consisting of altogether six configurations, both UO2 and mixed oxide, measured at two different temperatures at Studsvik. These configurations were selected for the purpose of the OECD/NEA uncertainty analysis in modeling benchmark activities. Uncertainties due to input data uncertainties, modeling errors, and numerical approximations were studied, with particular emphasis on the uncertainties in the nuclear cross-section data. The SUSD3D sensitivity-uncertainty code with the SCALE-6.0, JENDL-4, and/or JENDL-3.2 covariance data were used in this study. The consistency among the calculated-to-experiment values and the overall computational uncertainties is discussed.