It is often assumed that a binary begins to interact when one of its components makes contact with its Roche lobe, thus “switching on” a new evolutionary process. The example of Y Cygni is used to illustrate the view that the whole lifetime of a binary helps to determine whether or not its components will interact. Of particular importance is the interval between the formation of a binary and the arrival of its components on the main sequence, during which probably all binaries are interacting. Barring accidents, the properties of the components when they reach the main sequence will define the whole subsequent history of the system, including whether or not there will be subsequent phases of interaction triggered by contact with the Roche lobe. Like any other mechanical system a binary will tend towards the state of lowest energy consistent with the constraints on it. This it can do by losing mass, equalizing the component masses, or reducing its separation. We therefore expect systems to tend to small masses to mass-ratios of unity, or to coalesce into single stars. In any given system, probably all three tendencies exist, but one dominates. For example, W Ursae Majoris systems may be fusing into single stars. The rotation, chemical composition, and magnetic fields of the component stars may modify the evolution of a binary and be responsible for the variety of interacting systems that we observe. Most interacting pairs are losing mass to the interstellar medium, so a complete study of binary evolution must consider not only the dynamical, but also the chemical, effects of binary systems on the evolution of the Galaxy.