The authors present a framework for computing and evaluating linear projections of macro variables conditional on hypothetical paths of monetary policy. A modest policy intervention is a change in policy that does not significantly shift agents' beliefs about policy regime and does not generate quantitatively important expectations-formation effects of the kind Lucas (1976) emphasizes. The framework is applied to an econometric model of U.S. postwar monetary policy behavior. It finds that a rich class of interventions routinely considered by the Federal Reserve are modest and their impacts can be reliably forecast by an accurately identified linear model. Moreover, modest interventions can matter: They may shift the projected paths and probability distributions of macro variables in economically meaningful ways.