It continues to be heavily disputed whether foreign aid promotes economic growth in developing countries. In most cross-country regressions, aid is considered effective only if it shifts recipient countries to a significantly higher and sustainable growth path. We apply an alternative approach which is less demanding, based on the concept of temporary growth accelerations suggested by Hausmann, Pritchett and Rodrik. In assessing what can reasonably be expected from the donors' modest aid efforts, we do not only employ aggregate aid data but we also differentiate between major aid categories, including grants, loans and so-called short-impact aid. It turns out that aid flows have a small but significantly positive effect on the conditional probability of growth accelerations. This result holds across different estimation methods. Short-impact aid is found to be more effective in this respect, while we reject the view that grants are superior to loans. To the contrary, we find a stronger effect of loans. Furthermore, aid has become more effective during the second half of our sample. Typically, however, the significance of results crucially depends on the criteria applied to identify growth accelerations.