This paper attempts to explain the sources of real exchange rate fluctuations for a set of advanced economies and Central and Eastern European transition economies. To that end, we first estimate structural (identified) vector autoregression (SVAR) models, and decompose real and nominal exchange rate movements into those caused by real and nominal shocks. We then complete the previous step with an impulse-response analysis. There is evidence of instability in the variance decomposition of the real exchange rates for advanced economies across samples, with a growing importance of nominal shocks. Nominal shocks are also important in some transition economies.