A simple method was used to predict binding sites and to calculate the binding free energy for a xenon atom on a protein to determine the importance of the translational motion of water molecules in molecular recognition. We examined xenon bound on myoglobin and on a fragment of ammonium transporter. Despite the simplicity of our method, the predicted binding sites and the experimental results agree very well, and the estimated values of the free energy gain are also reasonable. We discuss the van der Waals picture of molecular recognition between a protein and a small hydrophobic molecule, such as an anesthetic molecule, which gives us a simple physical justification for the idea of a “lock and key” relationship used in conventional structure-based drug design programs.