We study information in the volatility of US Treasuries. We propose a no-arbitrage term structure model with a stochastic covariance of risks in the economy, and estimate it using high-frequency data and options. We identify volatilities of the expected short rate and of the term premium. Volatility of short rate expectations rises ahead of recessions and during stress in financial markets, while term premium volatility increases in the aftermath. Volatile short rate expectations predict economic activity independently of the term spread at horizons up to one year, and are related to measures of monetary policy uncertainty. The term premium volatility comoves with a more general level of economic policy uncertainty. We also study channels through which volatility affects model-based inference about the yield curve. A vast literature studies and decomposes the information contained in the nominal Treasury yield curve. Perspectives vary from latent factors, reduced-form macro-finance models, through structural settings but most of the time they serve the objective of understanding the Treasury risk premia and short rate expectations. The majority of the recent term structure literature has tackled this and related questions assuming constant volatility. That focus can be justified with tractability and the known difficulties with the joint modeling of the first and second moments of yields. Consequently, relatively little is known about information embedded in the Treasury market volatility. As the largest and the most liquid debt market in the world, US Treasuries provide investors, central banks and governments worldwide not only with the store of value and liquidity but also serve as the main source of collateral in various transactions. The fluctuations in Treasury yields and in their volatility have impact on a range of asset markets. A rise in interest rate volatility may signal macroeconomic uncertainty, but might also lead to an increase in haircuts for Treasury bonds used as collateral in about half of the repo transactions, and thus diminish the lending capacity in the financial system. 1 These considerations make the understanding of interest rate volatility especially compelling. In this paper, we ask what we can learn about the term structure and the economy by studying the time-varying second moments of the Treasury yield curve. Relative to the previous literature, our approach is comprised of several new elements: First, to identify the properties of interest rate volatility, we rely on nearly 20 years worth of high-frequency data on transactions in the US Treasury market. The dataset allows us to construct the realized covariance matrix of zero-coupon yields across different maturities. Additionally, we augment the realized co-volatility estimates with information from Treasury derivatives, covering risk-adjusted market expectations of volatility at several points along the yield curve. Second, we propose a no-arbitrage term structure model that is able to accommodate the multivariate dynamics of y...