This paper examines the methods used by the Bank of Japan for the estimation of the output gap. Attention is paid to the real-time estimation problem. After reviewing the evolution of output gap estimation at the Bank, I discuss advantages and disadvantages of the various output gap measures. First, I examine the usefulness of the output gap for inflation forecasting and show that the real-time output gap sometimes includes much noise. Second, I investigate the implications of the real-time estimation problem for the policy evaluation. Third, I exploit TANKAN to enhance the usefulness of the real-time output gap.