We extend the growth-at-risk (GaR) literature by examining US growth risks over 130 years using a time-varying parameter stochastic volatility regression model. This model effectively captures the distribution of GDP growth over long samples, accommodating changing relationships across variables and structural breaks. Our analysis offers several key insights for policymakers. We identify significant temporal variation in both the level and determinants of GaR. The stability of upside risks to GDP growth, as seen in previous research, is largely confined to the Great Moderation period, with a more balanced risk distribution prior to the 1970s. Additionally, the distribution of GDP growth has narrowed significantly since the end of the Bretton Woods system. Financial stress is consistently associated with higher downside risks, without affecting upside risks. Moreover, indicators such as credit growth and house prices influence both downside and upside risks during economic booms. Our findings also contribute to the financial cycle literature by providing a comprehensive view of the drivers and risks associated with economic booms and recessions over time.