The MM/PBSA and MM/GBSA methods are widely used to estimate ligand-binding affinities for small ligands to biological macromolecules. These methods combine molecular mechanics with continuum solvation models (Poisson-Boltzmann or generalized Born) and are based on molecular dynamics simulations. They offer a balance between accuracy and computational cost, falling between empirical scoring and alchemical perturbation methods. However, they have limitations, including approximations such as neglecting conformational entropy and water molecule effects in binding sites. The methods involve several terms: electrostatics, polar and non-polar solvation, van der Waals, and entropy. The results depend on factors like the continuum-solvation model, charges, dielectric constant, and sampling methods. While the methods can be improved with more accurate approaches like quantum mechanics or polarizable force fields, these often do not consistently enhance results. The MM/PBSA method is modular and does not require training sets, making it useful for virtual screening and docking. However, it is not accurate enough for predictive drug design. The method's precision is limited by the number of simulations and the statistical uncertainty of the results. The entropy term is particularly challenging, as it is calculated using normal-mode analysis and can introduce large uncertainties. The non-polar solvation term is often small and has limited impact on the results. The electrostatic term depends on the charges used for the ligand and receptor, and various force fields have been tested. The dielectric constant also plays a role in the results. Overall, the MM/PBSA method is a popular but not fully accurate approach for estimating ligand-binding affinities.The MM/PBSA and MM/GBSA methods are widely used to estimate ligand-binding affinities for small ligands to biological macromolecules. These methods combine molecular mechanics with continuum solvation models (Poisson-Boltzmann or generalized Born) and are based on molecular dynamics simulations. They offer a balance between accuracy and computational cost, falling between empirical scoring and alchemical perturbation methods. However, they have limitations, including approximations such as neglecting conformational entropy and water molecule effects in binding sites. The methods involve several terms: electrostatics, polar and non-polar solvation, van der Waals, and entropy. The results depend on factors like the continuum-solvation model, charges, dielectric constant, and sampling methods. While the methods can be improved with more accurate approaches like quantum mechanics or polarizable force fields, these often do not consistently enhance results. The MM/PBSA method is modular and does not require training sets, making it useful for virtual screening and docking. However, it is not accurate enough for predictive drug design. The method's precision is limited by the number of simulations and the statistical uncertainty of the results. The entropy term is particularly challenging, as it is calculated using normal-mode analysis and can introduce large uncertainties. The non-polar solvation term is often small and has limited impact on the results. The electrostatic term depends on the charges used for the ligand and receptor, and various force fields have been tested. The dielectric constant also plays a role in the results. Overall, the MM/PBSA method is a popular but not fully accurate approach for estimating ligand-binding affinities.