We study the sources of the Great Moderation by estimating a variety of medium-scale DSGE models that incorporate regime switches in shock variances and in the inflation target. The best-fit model, the one with two regimes in shock variances, gives quantitatively different dynamics in comparison with the benchmark constant-parameter model. Our estimates show that three kinds of shocks accounted for most of the Great Moderation and business-cycle fluctuations: capital depreciation shocks, neutral technology shocks, and wage markup shocks. In contrast to the existing literature, we find that changes in the inflation target or shocks in the investment-specific technology played little role in macroeconomic volatility. Moreover, our estimates indicate much less nominal rigidities than those suggested in the literature.