Recent theoretical and empirical research in international macroeconomics has rediscovered the problem of purchasing power parity (PPP). Empirically, PPP is a bad approximation of both the short-term and medium-term properties of the data. Economists have had difficulties in explaining the persistent misalignments of real exchange rates, but new empirical research by Clarida and Galí (1995) suggests that much of these real exchange rate movements are due to relative demand shocks. The present paper challenges this view by using an extended version of their structural vector autoregressive (SVAR) model in order to identify a larger number of real shocks (labour supply, productivity and aggregate demand) and nominal shocks (money demand and money supply). It is found that whilst some of their results go through in our extended framework, there is serious doubt with respect to the appropriateness of labelling those shocks which drive real exchange rates as aggregate demand disturbances.