[en] The SLAROM code, performing fast reactor cell calculation based on a deterministic methodology, has been revised by adding the universal module PEACO of generating Ultra-fine group neutron spectra. The revised SLAROM, then, was utilized for evaluating reaction rate distributions in ZPPR-13A simulated by a 2-dim RZ homogeneous model, although actually ZPPR-13A composed of radially heterogeneous cells. The reaction rate distributions of ZPPR-13A were also calculated by the code MVP, that is a continuous energy Monte Carlo calculation code based on a probabilistic methodology. By comparing both results, it was concluded that the module PEACO has excellent capability for evaluating highly accurate effective cross sections. Also it was proved that the use of a new fine group cross section library set (next generation set), reflecting behavior of cross sections of structural materials, such as Fe and 0, in the fast neutron energy region, is indispensable for attaining a better agreement within 1% between both calculation methods. Also, for production of a next generation set of group cross sections, the code NJOY97.V107 was added to the group cross section production system and both front and end processing parts were prepared. This system was utilized to produce the new 70 group JFS-3 library using the evaluated nuclear data library JENDL-3.2. Furthermore, to confirm the capability of this new group cross section production system, the above new JFS-3 library was applied to core performance analysis of ZPPR-9 core with a 2-dim RZ homogeneous model and analysis of heterogeneous cells of ZPPR-9 core by using the deterministic method. Also the analysis using the code MVP was performed. By comparison of both results the following conclusion has been derived; the deterministic method, with the PEACO module for resonance cross sections, contributes to improve accuracy of predicting reaction rate distributions and Na void reactivity in fast reactor cores. And it becomes clear that the new JFS-3 library results in the almost same performance as those will the JSF-2-J3.2 library utilized so far. (author)