Modern economic theory considers expectations as a key determinant of actual inflation. How agents form those expectations therefore plays a central role in macroeconomic dynamics and policy-making. The understanding of the expectation formation process and the real-time estimation of expectations are especially important for central banks because they need to be sure that longer-term inflation expectations are anchored at the target of inflation, set by the central bank. When expectations are anchored — it is a clear sign that the monetary policy is effective and that markets trust the central bank. However, it is not easy to assess the expected inflation: it is not observable and cannot be directly measured. Central banks can only use the indirect estimates of this variable. For many years the main theoretical framework for modeling and analysis of inflation expectations was Phillips curve with rational expectations which substituted the adaptive expectations. Today many alternative models of expectation formation are available. The article provides a brief overview of the evolution of theoretical approaches to inflation expectation formation and their impact on the monetary policy. Besides, using the experience of the U.S., the article addresses two main ways to gauge inflation expectations empirically — survey-based measures (for different groups of respondents) and measures based on the data from American financial markets. Shortcomings and merits of both approaches are discussed, as well as the importance of highly developed financial markets, which can become the source of more precise information on inflation expectations.