That biological features may change their function during evolution has long been recognized. Particularly, the acquisition of new functions by molecules involved in developmental pathways is suspected to cause important morphologic novelties. However, the current terminology describing functional changes during evolution (co-option or recruitment) fails to recognize important biologic distinctions between diverse evolutionary routes involving functional shifts. The main goal of our work is to stress the importance of an apparently trivial distinction: Whether or not the element that adopts a new function (anything from a morphologic structure to a protein domain) is a single or a duplicated element. We propose that natural selection must act in a radically different way, depending on the historic succession of co-option and duplication events; that is, co-option may provide the selective pressure for a subsequent gene duplication or could be a stabilizing factor that helps maintain redundancy after gene duplication. We review the evidence available on functional changes, focusing whenever possible on developmental molecules, and we propose a conceptual framework for the study of functional shifts during evolution with a level of resolution appropriate to the power of our current methodologies.