Noble metal-contained multi-principal element alloys (MPEAs), primarily composed of less noble elements, emerge as promising catalysts by enhancing catalytic performance while reducing cost compared with conventional noble metal catalysts. However, limited understanding of the underlying catalytic mechanisms has hindered the exploration of practical applications of MPEA catalysts due to the complex atomic environments, unique surface structures, and synergistic effects associated with MPEAs. In this work, the oxygen evolution reaction (OER) performance of FCC_CoCuFeNiPd and FCC_CoCuFeNiRu was theoretically investigated in detail under alkaline conditions, based on the sublattice preference of constituent atoms and the site preference of intermediates. For this purpose, firstly, the site occupying fractions (SOFs) of the constituent elements on the sublattices were predicted by combining the sublattice model with computational thermodynamics. Then, the reaction surfaces were cleaved from the bulk alloy models, which allowed a richer set of adsorption and reaction sites to be systematically considered. This approach contrasted with the commonly used, but oversimplified, special quasi-random structure (SQS) model, in which all constituent alloying elements were assumed to be perfectly randomly distributed on different sublattices, i.e., the vertex positions (1a) and face-centered positions (3c) of the ordered cubic crystal lattice. Based on the advanced model, we demonstrated that CoCuFeNiRu outperformed CoCuFeNiPd, exhibiting a lower average overpotential (0.210 V vs. 0.325 V) and a more favorable optimal surface overpotential (0.181 V vs. 0.214 V). This was determined by calculating the Gibbs free energies of *OH, *O, and *OOH intermediates individually at 91 adsorption sites on the Slab surface. Ru exhibited adsorption free energies of intermediates closer to ideal values, facilitating reaction steps, whereas Pd’s higher free energies raised the overall overpotential. Meanwhile, we found that the average overpotential of CoCuFeNiPd based on the SOFs model (0.214 V) aligned better with the experimental value (0.193 V) than that predicted by the SQS model (0.25 V), highlighting the necessity and importance of considering the sublattice preferences of constituent alloying elements. Our contribution offers a theoretical foundation to explore and design high-performance noble metal-contained MPEA catalysts.