Existing results on fiscal federalism and economic growth are ambiguous. * This paper studies this relation using a Bayesian model averaging approach (BMA). * BMA allows for avoiding biases due to over restrictive model specification. * The analysis is conducted for a panel of 23 OECD countries from 1975 to 2000. * There is no robust positive or negative link between fiscal federalism and growth. Following the ambiguous results in the literature aimed at understanding the empirical link between fiscal federalism and economic growth, this paper revisits the question using a Bayesian model averaging approach. The analysis suggests that the failure to appropriately account for model uncertainty may have previously led to biased estimates. The results from a sample of 23 OECD countries over 1975-2000 indicate that after controlling for unobserved country heterogeneity, there is no robust link, neither positive, nor negative, between output growth and fiscal federalism. Clearly, widely recognized issues of endogeneity and causality that are typical to the empirical growth literature in general remain unresolved.