In this paper we revisit the contribution of financial development on growth through panel data regressions based on a panel set covering 82 countries and six five-year periods of time over the years 1960 to 1990. Our results do not confirm the existence of a positive relation between financial development and growth, be it through savings mobilisation or through the quality of its allocation. We resolve this paradox by interpreting the role of financial sector in a multiple equilibra model which generates threshold effects. Such threshold effects are observable in the country fix effects of our regressions.