We underline that some results obtained in the literature on central bank transparency may be quite different when we take a non-Bayesian approach to uncertainty, where "ambiguity" is taken into account. We consider some specific argument of the literature (obtained under a Bayesian approach), which implies that political uncertainty can be beneficial and central bank transparency harmful. We show that when ambiguity is large enough, these results do not hold anymore: political uncertainty is always harmful and central bank transparency always beneficial. Furthermore, as soon as we depart from the Bayesian case, Knightian uncertainty is always harmful. JEL Classification: E58; E52 Keywords: central bank transparency; political uncertainty; ...