Theoretical reflections about System Dynamics (SD) have usually been grounded in the developments of what can be called “general philosophy of science”. In our paper, a philosophical approach more sensitive to the peculiarities of SD is proposed that is closely linked to the recent constructivist proposal of John Searle and to the expressivist theses of Robert Brandom. We will focus on three very important conceptual problems –the ontological problem of realism concerning the structures postulated by SD models, the epistemological problem of the explanatory power of SD models, and the methodological charge of merely producing a kind of “patchwork” when building of SD models--, arguing that by combining the constructivist and expressivist philosophical perspectives of those authors in a certain way would offer a better understanding of scientific and technical activities such as SD modelling.