Metal nanoparticles are widely used as heterogeneous catalysts, and their catalytic performance is governed by multiple factors, including reactive sites (facets, edges, and corners), particle size, interactions with supports, and the formation of new active sites such as perimeter sites, surface defects, and dopant-induced sites. Because these factors are intricately intertwined, a microscopic understanding of their roles is essential for rational catalyst design. Computational chemistry provides a powerful framework for systematically exploring reaction pathways and mechanisms under well-defined model conditions, where appropriate simplification is necessary to maintain feasible computational cost. In this article, we review our theoretical studies based mainly on density functional theory, focusing on how cluster size, reaction sites, external fields, dopants, and metal–support interactions influence catalytic reactivity. By integrating theoretical analysis with experimental findings, we demonstrate how this combined approach provides detailed insight into nanoscale elementary processes and practical guidelines for catalyst design.