The debate concerning the various determinants of economic growth has attracted considerable attention. The argument according to which endogenous growth models explain long-term economic growth is often put forward. Particularly, it is held that the production of knowledge by education induces self-sustained growth. In spite of numerous theoretical developments, attempts at empirical verification give contradictory conclusions. The aims of this article are therefore to undertake a critical reading of the theoretical contribution of new growth theories and to present an empirical testing for France in the nineteenth and twentieth centuries to justify or invalidate the probable endogenous nature of economic growth induced by education. The results are surprising and in contradiction with the hypothesis of new growth theories: human capital returns are decreasing and knowledge produced by education cannot be the engine of self-maintained economic growth. These results induce new kinds of questions: does that mean that there is no causality link between education and economic growth? Econometric tests seems to invalidate this causality relation.