This paper employs linear and nonlinear ARDL models to examine the short-run and long-run relationship between stock prices and exchange rates in the G7 countries. Both the flow-oriented approach that exchange rates affect stock prices and the portfolio balance approach that stock prices affect exchange rates are supported in the short-run. Neither model is supported in the long-run using linear ARDL models, but the nonlinear ARDL model shows evidence supporting the portfolio balance approach in four of the countries. In these four countries we find that rising and falling stock prices have significant long-run effects on their exchange rates. Furthermore, Granger causality tests confirm that causality runs from stock prices to exchange rates in six of the countries. Thus, the use of a longer and more recent data set provides stronger long-run support for the portfolio balance approach than found in most of the recent literature, while we confirm results of recent research showing no long-run evidence of causation running from exchange rates to stock prices.