This paper empirically evaluates the predictive performance of the International Monetary Fund's (IMF) exchange rate assessments with respect to future exchange rate movements. The assessments of real trade-weighted exchange rates were conducted from 2006 to 2011, and were based on three state-of-the-art exchange rate models with a medium-term focus which were developed by the IMF. The empirical analysis using 26 advanced and emerging market economy currencies reveals that the "diagnosis" of undervalued or overvalued currencies based on these models has significant predictive power with respect to future exchange rate movements, with one model outperforming the other two. The models are better at predicting future exchange rate movements in advanced and open economies. Controlling for the exchange rate regime does not increase the predictive power of the assessments. Furthermore, the directional accuracy of the IMF assessments is found to be higher than market expectations.