We apply the trading model of Fleming et al (1998). to a number of currency markets. The model posits that two markets can have common volatility structures as a result of receiving common information and from cross-hedging activity where a position in one currency is used to hedge risk in a position taken in another. Our results imply that the model is effective in identifying common information flows and volatility spillovers in the currency markets and that some of these effects are lost when simply examining raw correlations. A series of specification tests of the 21 bivariate systems that are examined provides support for the trading model in the foreign exchange context.