A multitude of biological processes requires the participation of specific cations, such as H + , Na + , K + , Ca 2+ , and Mg 2+ . Many of these processes can take place only when proteins have the ability to discriminate between different ions with a very high fidelity. How this is possible is a fundamental question that has fascinated scientists for a long time. At the most fundamental level, it is anticipated that ion selectivity must result from a delicate balance of strong interactions. Yet, identifying and quantifying the key microscopic factors is difficult, as many of these cannot be directly measured by experiments. Theory and computations can contribute by providing a virtual route to complement the missing information. Because ion selectivity is often dominated by thermodynamic factors, detailed molecular dynamics (MD) free energy simulations become important tools. This was vividly illustrated early on with studies of ion solvation (Straatsma and Berendsen, 1988;Åqvist, 1990) and ion-selective systems (Lybrand et al., 1986;Grootenhuis and Kollman, 1989;Åqvist, 1992). These pioneering studies inspired our own efforts.In this Perspective, we aim to present our understanding of ion selectivity as it has evolved over approximately 15 years from studies based on various specific structures: gramicidin A channels (e.g.