In risk management, foreign investors or multinational corporations are highly interested in knowing how volatile acurrency is in order to hedge risk. In this paper, using daily exchange rates and the Exponential Weighted Moving Average (EWMA) model, we perform volatility forecasting. We will investigate how used available time series affect the forecastings, i.e. how reliable our forecasting is depending on the period of used available data. We will also experiment the effects of the decay factor appearing in the model used on the forecasts.The results show that for the data used, it is optimal to use a smaller value of the decay factor and also for longer out-of-sample periods the forecasts get closer to the reality.