This paper compares the GARCH in Mean, the GARCH and the EGARCH models in measuring exchange rate volatility to determine which model is more efficient in terms of forecasting of volatility. Analysis of forecasts of exchange rate volatility using Mean Squared Error (MSE) shows that the EGARCH (l, l) model does better in describing the data for half of the sample countries' exchange rates than the GARCH (1,1) and the GARCH-M (1,1) models. When the Mean Absolute Percentage Error (MAPE) is used for performance measure, the results are mixed. These results imply that the GARCH (l, 1) model might not be an excellent model for measuring and forecasting volatility when it varies over time.