Although there are simple and usual arguments claiming that reserves might offer protection at sudden stops, few papers about this subject are found in the literature. The goal of this study is to try to identify and quantify the real benefits (or costs) of having reserves at this kind of crisis. We begin with the construction of a model that, contrary to what is usually done, allows the stock of reserves to impact the dynamics of capital flows, through expectations. It suggests that reserves reduce the probability of occurrence of crisis, but, once it is starts, it tends to be stronger. We therefore study these relations empirically, with a panel of countries. We don´t find effects upon the probability of sudden stop, but, indeed, the regressions suggest that reserves tend to increase the fall in capital flows at crises. Nevertheless, we find no evidence that this instrument affects the cost of the sudden stop (measured by the product fall).