This study examines on the basis of economic theory the determinants of exchange rate volatilities for a large number of currencies. We relate daily changes in GARCH(1,1) volatilities of exchange rates to the volatility changes of several of their presumed fundamental economic determinants in the context of a portfolio balance model. The use of high-frequency data limits the choice of the explanatory economic variables that can be included in empirical estimates. The first differences of GARCH(1,1) volatilities of share and bond price indices reflect portfolio trading decisions in corresponding markets for both assets. In the same vein, first differences of the gold price volatility, as an additional determinant, are related to exchange rate volatilities of two commodity currencies in the sample. The panel data estimates, using the Seemingly Unrelated Regression technique, produce coefficients with the expected signs and statistical significance. The results of our study enhance our understanding of high-frequency currency volatility changes for 19 currencies beyond the purview of announcement effects in the event studies framework.