This paper nowcasts the Euro-Dollar short-run exchange rate by using a MF-TVP-FAVAR model. The FAVAR framework improves forecasting accuracy by expanding the information set of the previously widely used VAR models. We adopt a flexible modelling approach that adjusts for structural breaks in the data and money demand instability; it also prevents information loss due to variables being quoted at mixed frequencies. We estimate our model by using a dual conditionality linear Kalman filtering/smoothing. Our results indicate that the specified model outperforms the Random Walk and other structural models at all forecasting horizons.