Empirical research on the microeconomics of currency markets, an area known sometimes as 'currency market microstructure,' has taken off in the past decade. This paper extracts from this research four lessons for modelling short-run exchange-rate dynamics. First, currency flows are a key determinant of exchange rates, and models should have these flows fully in focus. The appropriate equilibrium condition may be flow-currency supply equals flow-currency demand. Second, the flows of first-order relevance include those of 'financial' traders, who use currencies essentially as a store of value, and 'commercial' traders, who use currencies as a medium of exchange. Financial flows and commercial flows should be negatively related to each other, meaning that financial demand tends to be met by commercial supply. Also, financial flows should be positively related to exchange rates. Third, financial traders are motivated by profits, rather than consumption, and behave as if risk averse. Fourth, commercial traders are motivated by exchange-rate levels and rationally choose not to speculate. The workhorse models of international macroeconomics do not fit most of these lessons, and these important lacunae in their microfoundations may help explain their limited empirical success. The paper sketches an optimizing model of currency flows that is consistent with the lessons and has an encouraging empirical record. Copyright © 2006 John Wiley & Sons, Ltd.