We examine the relationship between ice conditions in the Baltic Sea and the large-scale atmospheric circulation patterns that create them. Singular Spectrum Analysis (SSA) is used to extract long-term trends and quasi-regular oscillations from time series of winter air temperature, date of ice break-up, maximum annual ice extent in the Baltic Sea (BMI) since 1708 and seasonal North Atlantic Oscillation (NAO) and Arctic Oscillation (AO) indices. The AO exhibits power at periods of 2.3 and 13.9 years as well as at 7.8 years, and in general the NAO can be viewed as a subset of the AO as regards the Baltic region. Time series of ice break-up date also reflects variations in winter AO power in the 13.9-year power band, but not in the NAO 7.8-year band. The BMI seems to act as a non-linear filter on the fairly weak climate oscillations, increasing the signal-to-noise ratio of the oscillations significantly. ¿ 2001 American Geophysical Union