Over the past few years, the ability of the United States to finance its current account deficit has been facilitated by massive purchases of U.S. Treasury bonds and agency securities by Asian central banks. In this process, Asian central banks have accumulated large stockpiles of U.S.-dollar foreign exchange reserves. How far is the current level of reserves from that predicted by the standard macroeconomic determinants? The authors answer this question by using Pedroni's (1999) panel cointegration tests as the basis for the estimation of a long-run reserve-demand function in a panel of eight Asian emerging-market economies. This is a key innovation relative to the existing research on international reserves modelling: although the data are typically I(1), the literature ignores this fact and makes statistical inference based on unadjusted standard errors. While the authors find evidence of a positive structural break in the demand for international reserves by Asian central banks in the aftermath of the financial crisis of 1997–98, their results indicate that the actual level of reserves accumulated in 2003–04 was still in excess relative to that predicted by the model. Therefore, as long as historical relationships hold, a slowdown in the rate of accumulation of reserves is likely. This poses negative risks for the U.S. dollar. However, both the substantial capital losses that Asian central banks would incur if they were to drastically change their holding policy and the evidence that the currency composition of reserves evolves only gradually mitigate the risks of a rapid depreciation of the U.S. dollar triggered by Asian central banks.