There is compelling evidence that the number of cortical areas is variable across species, ranging from the order of 10 to 20 in species with little neocortex to perhaps a hundred or more in humans. The first mammals had few areas, and more fields evolved in several lines of descent. There are several general possibilities for how new cortical areas emerged in evolution, including the gradual differentiation of one area from another, the addition of new cortical tissue that became invaded by axons from other structures to create new fields, and the sudden duplication of existing areas as a result of mutation. A specific version of a "gradual differentiation" theory is presented. New areas might emerge as a result of a process in which classes of inputs to an area first segregate into modular groups of neurons, and then these modules coalesce to ultimately form totally separate populations. All stages allow expected functions to be mediated, and there might be different functional advantages for each stage. Thus, stages may be stable and change is not inevitable. As a result, all stages are represented in the organizations of existing fields of extant mammals.