The rapid progression of computer technology, encompassing both software algorithms and hardware, has significantly sped up the process of virtual screening, markedly enhancing the identification of potent small molecules for drug development. The use of computational modeling to understand the interactions between RNAs and the interdependency of RNAs with other molecules have become a critical component in the search for RNA-targeted therapeutics. Presently, the primary approach to modeling RNA–ligand interactions relies on the docking-and-scoring technique. For docking and scoring to be precise, it must address four essential challenges: (1) the ligand’s conformational flexibility, (2) the RNA’s conformational flexibility, (3) effective exploration of binding sites and poses, and (4) precise evaluation of various binding configurations. Furthermore, the task of modeling ligand binding to RNA, a negatively charged macromolecule, entails additional complexities over protein-ligand docking, such as the influence of metal ions. Thermodynamic approaches, employing both physics-based and knowledge-based evaluation functions, have demonstrated promising outcomes in accurately predicting ligand binding poses and affinity. Kinetic modeling of ligand binding, introducing the concept of dissociation kinetics into the docking process, has been shown to enhance the prediction of in vivo drug effectiveness. The emergence of deep learning techniques has introduced novel predictive tools for RNA–small molecule interactions. This section provides a summary of the latest computational strategies for RNA–ligand docking, highlighting their strengths and limitations.