In this essay, I develop a comparative analysis of the role of prediction in ideal concepts within Natural Sciences, and its function in Weber’s ideal-types. I will argue for the predictive capacity of the former and against of the latter. A conceptual analysis of the historical debate Erklären-Verstehen will be also presented. I will defend that our issue can adequately be traced back to that discussion. Finally, a typology of the notion of prediction will be exposed to open the way towards a redefinition of the predictive capacity of ideal-types.