The textbook J-curve, which describes the trade balance of a nation initially worsening after a real depreciation of the home currency, then getting better, has drawn increased attention in the past decade. The decline in the exchange value of the dollar and the persistence of the American trade deficit during the 1980s prompted economists to re-examine the traditional theory. A number of studies [e.g., Rose and Yellen, JME, 1989] question the empirical validity of the theory for the U.S. A necessary condition for the J-curve to obtain is that a nation's trade balance respond to changes in the real exchange rate. This study tests that hypothesis by examining the causality (in the Granger sense) of the deutsche mark and West Germany's trade balance during the current floating exchange rate regimes. Germany experienced a persistent trade surplus during the 1970s that seemed to react little to the exchange rate, providing a natural experiment. It is well-known that regressing I(1) variables on one another leads to spurious regression problems. Augmented Dickey-Fuller tests reject the null hypothesis of a unit root for the real effective exchange rate but not for the real per capita trade balance. However, economic theory suggests that the real trade balance cannot be unbounded. When tested with linear and quadratic trend terms [suggested in Ouliaris et al., Advances in Econometrics, 1989], the null is not rejected nearly as strongly. It is felt that this non-rejection is due to the small sample size and low power of the tests. The remainder of this paper treats the real per capita trade balance as trend stationary. The Granger-causality regression is then written as follows: