In this paper, the authors investigate the importance for the monetary model of exchange rate determination of J. Frankel (1979) of allowing unrestricted dynamics and of accounting for shifts in the demand for money due to financial innovation. Based on estimation and simulation results for the Canada-U.S. bilateral exchange rate over the 1970-85 period, the aut hors find strong evidence in favor of a generalized approach to dynamic specification, but find that the shif t adjustment of official money supply data has only minor implications for the empirical results. Moreover, neither modification is sufficient to generate a monetary model that is well supported by the data.