Measured by transaction volume, foreign exchange swaps are the largest market in the world. However, there are very few empirical studies of swap rates. Theoretically, covered interest parity is commonly assumed. But what factors determine arbitrage opportunities? We create a unique microstructure model of exchange rate activity to identify theoretical predictions regarding covered interest arbitrage in a market with a dominant market maker. Using a unique data set of actual, recorded swap transactions, not price quotes, the model is verified as we find economically significant returns that depend in part on market volatility, contract irregularity and trader identity. (C) 2012 Elsevier B. V. All rights reserved.