This article presents a general classification of key exchangerate forecasting models and methods. An overview of forecast centers, their forecasting models and techniques as well as forecasts is also given. With a view to evaluate forecasts accuracy mean absolute percentage errors (MAPE) were calculated and compared between different forecast centers. Despite that most of the exchangerate forecasts have a much lower error than commodities prices forecasts; they are still not accurate enough (with MAPE about 57 %). According to the evaluation the most accurate forecasts are based on structural models, conditional assumptions (about constant exchange rates) and consensus forecasts. Longer horizons can improve forecast accuracy not significantly, while forecasts made for 2008-2009 crisis years usually deviate strongly from the fact.