This paper confronts three conundrums. First, does the relationship between aid and growth fade over time when aid is successful? Second, why are aid inflows neglected in the literature on growth acceleration (or episodes). Third, why is country vulnerability overlooked in the same literature? Our purpose is to address these puzzles, and in doing so two hypotheses are formulated and tested. First, we assume that aid can have a positive (catalytic) effect on the launching of growth episodes, as well as on their duration. Second, we assume that this effect is all the more significant with the intensity of the exogeneous shocks the country faces. Econometric tests do not reject these hypotheses. The paper first considers the origin of the puzzles and explains the hypotheses presented as the answer, and then introduces the models used to test these. Finally, it assesses the results and their implications. Once again, it appears that vulnerability does matter with regard to the impact of aid on both the probability of an occurrence of growth spells and on their duration.