The numerical domain is unique in at least three ways. First, numbers represent a particular aspect of reality (i.e., numerosity). Second, numbers are the object of specific processing such as calculation, parity judgment, magnitude comparison, and so on. And third, numbers can be represented in different formats: Arabic numbers (i.e., digit strings, e.g., 25), written numbers (i.e., letter strings; e.g., twenty-five) or spoken numbers (i.e., sequences of sounds; e.g., TWENTY- FIVE), Roman numerals (XXV), and so on'. These characteristics have raised specific questions. Four of them will be considered in this chapter. First, are there functionally independent systems to process the different formats for numbers? Second, how does the human subject translate a number presented in one format into another format? Third, what are the basic functional components for calculation? And fourth, in which internal code(s) are arithmetic facts stored in memory?