Observational data used to evaluate quantitatively the relationships between arctic sea ice extent and the high-latitude atmospheric circulation on the seasonal time scale. The sea ice data set contains 300 monthly grids of observed sea ice concentrations. The atmospheric variables include sea level pressure, surface termperature, 700-mbar height, and 700-mbar temperature. Statistically significant correlations between the dominant modes of atmospheric and sea ice variability are found at atmospheric lags of up to 2 months and ice lags of up to 4 months. The surface temperature field generally shows the strongest relationship to the sea ice fluctuations. The strongest correlations between ice anomalies and subsequent atmospheric fluctuations are found in the autumn months of increasing ice extent. Evidence of ice-atmosphere coupling is also found in the mid-latitude fields of the North Atlantic. The meteorological difference fields derived from years of extreme ice extent contain statistically significant pressure differences of up to 10--15 mbar, surface temperature differences of up to 8 ¿--9 ¿K, and 700-mbar height differences of up to 16--18 decameters. The anomaly centers tend to migrate seasonally with the ice edge. The statistical predictability of large-scale sea ice fluctuations decays to the level of no skill at a forecast interval of 5--6 months.