Here I describe two studies. The first is a longitudinal study of arithmetical development in 159 children from kindergarten to age 11. We found that a very simple test of numerosity processing in kindergarten predicted age-appropriate arithmetical performance to age 11, in both accuracy and strategy. In a separate study of 260 twins between 8 and 14 years. We found that numerosity processing performance correlated with grey matter density in the area of the parietallobe. Both the behavioural capacity and the underlying brain structure werecorrelated. In a Cross-Twin Cross-Trait genetic analysis we found that the relationship between the ability to enumerate dot displays and simple arithmetic shows is highly heritable, suggesting that the basic capacity to estimate numerosity is closely tied to the ability tolearn arithmetic. I discuss the theoretical and practical implications of these findings.