Recent research stresses the importance of supportive initial conditions for countries to reap the benefits of financial integration. This paper revisits the robustness and relative importance of different thresholds in the linkbetween financial openness and growth. It offers a comprehensive analysis of threshold effects in a common empirical framework. We employ Bayesian Model Averaging (BMA) techniques to appropriately account for the problem of model uncertainty in cross-country growth regressions which has been previously neglected by the financial openness literature. Model uncertainty arises along several dimensions: First, there is uncertainty about the control variables to include given the plethora of suggested growth determinants. Further, and more specific to the context analyzed, is the uncertainty about the existence of thresholds and most importantly uncertainty about the nature of the threshold given the long list of preconditions proposed in the literature. We find robust evidence of threshold effects. Specifically, our results suggest that FDI inflows are only beneficial if certain preconditions are met while debt flows might reduce growth under unfavorable initial conditions. After explicitly accounting for uncertainty with respect to the nature of the threshold, we find institutional thresholds to dominate other potential thresholds. In particular, we uncover strongest evidence that FDI inflows are positively correlated with long term growth in countries characterized by low levels of corruption, while a combination of supportive political and property rights institutions are crucial to avert the risks of debt flows.