Are empirical measures of uncertainty informative about risks to future economic activity? I use quantile regression analysis and density predictions on United States data to show that the relationship between macroeconomic uncertainty and future GDP growth is nonlinear and asymmetric. The left tail of the distribution of future GDP growth is highly responsive to fluctuations in macroeconomic uncertainty, whereas the right tail is relatively stable. As such, macroeconomic uncertainty predicts downside risks to growth but is less informative about upside risks. When combined with an index of financial conditions—a previously proposed predictor of downside risks to growth—macroeconomic uncertainty carries a larger weight in the optimal predictive density. Finally, I provide evidence that alternative empirical measures of uncertainty, such as economic policy uncertainty and geopolitical risk, do not predict risks to the economic outlook. These results hold for a larger sample of countries and underline the importance of differentiating between measures of uncertainty when predicting risks to growth.