Using a gravity equation, Andrew Rose finds that currency-union countries trade more intensively with each other than do other country pairs, and others report same result. Using a nonparametric test, however, Persson finds that trade flows between currency-union countries are not significantly different in size from other bilateral trade flows. Using another way to identify a currency-union country, this paper reproduces Persson's result but also produces an anomaly. When Rose's gravity equation is re-estimated using the data set furnished by the alternative definition of a currency-union country, Rose's result survives, although the currency-union effect is smaller than the effect he reported initially.