We estimate perceptions about the Fed’s monetary policy rule from micro data on professional forecasters. The perceived rule varies significantly over time, with important consequences for monetary policy and bond markets. Over the monetary policy cycle, easings are perceived to be quick and surprising, while tightenings are perceived to be gradual and data-dependent. Consistent with the idea that forecasters learn about the policy rule from policy decisions, the perceived monetary policy rule responds to high-frequency monetary policy surprises. Variation in the perceived rule impacts financial markets, explaining changes in the sensitivity of interest rates to macroeconomic announcements and affecting risk premia on long-term Treasury bonds. It also helps explain forecast errors for the future federal funds rate. We interpret these findings through the lens of a model with forecaster heterogeneity and learning from observed policy decisions.