Computational prediction of protein-ligand binding involves initial determination of the binding mode and subsequent evaluation of the strength of the protein-ligand interactions, which directly correlates with ligand binding affinities. As a consequence of increasing computer power, rigorous approaches to calculate protein-ligand binding affinities, such as free energy perturbation (FEP) methods, are becoming an essential part of the toolbox of computer-aided drug design. In this chapter, we provide a general overview of these methods and introduce the QFEP modules, which are open-source API workflows based on our molecular dynamics (MD) package Q. The module QligFEP allows estimation of relative binding affinities along ligand series, while QresFEP is a module to estimate binding affinity shifts caused by single-point mutations of the protein. We herein provide guidelines for the use of each of these modules based on data extracted from ligand-design projects. While these modules are stand-alone, the combined use of the two workflows in a drug-design project yields complementary perspectives of the ligand binding problem, providing two sides of the same coin. The selected case studies illustrate how to use QFEP to approach the two key questions associated with ligand binding prediction: identifying the most favorable binding mode from different alternatives and establishing structure-affinity relationships that allow the rational optimization of hit compounds.