In recent times, our healthcare system is being challenged by many drug-resistant microorganisms and ageing-associated diseases for which we do not have any drugs or drugs with poor therapeutic profile. With pharmaceutical technological advancements, increasing computational power and growth of related biomedical fields, there have been dramatic increase in the number of drugs approved in general, but still way behind in drug discovery for certain class of diseases. Now, we have access to bigger genomics database, better biophysical methods, and knowledge about chemical space with which we should be able to easily explore and predict synthetically feasible compounds for the lead optimization process. In this chapter, we discuss the limitations and highlights of currently available computational methods used for protein–ligand binding affinities estimation and this includes force-field, ab initio electronic structure theory and machine learning approaches. Since the electronic structure-based approach cannot be applied to systems of larger length scale, the free energy methods based on this employ certain approximations, and these have been discussed in detail in this chapter. Recently, the methods based on electronic structure theory and machine learning approaches also are successfully being used to compute protein–ligand binding affinities and other pharmacokinetic and pharmacodynamic properties and so have greater potential to take forward computer-aided drug discovery to newer heights.