Modeling and forecasting volatility in recent years has received considerable attention of academics, investors, but also the creators of economic policy. The reason is that volatility is an important indicator of the level of risk in financial markets. The aim of this study is the first detailed analysis of time series of income in foreign exchange rates EUR / CZK and then find the most suitable model and assess its predictive performance in predicting volatility analyzed revenues. For this purpose, use standard econometric models of apparatus, together with the conditional heteroskedasticity, namely GARCH model and its asymmetric variant GJR-GARCH with normal, Student's t-slanted and Student's t-distribution. Results evaluated GARCH (2,1) model as the best option for modeling the volatility observed yields.