This paper summarizes the long term and decadal time scale fluctuations of atmosphericcirculation in the Southern(Hemisphere)middle and high latitudes.The long time series ofAntarctic Oscillation Indices(AOIs)were established for January.April.July and October basedon the historical sea level pressure maps and NCEP/NCAR reanalysis surface pressure data set.Itwas found that the AOI of January had upward trend at a rate of 1.17 hPa/100 a.and July haddownward trend at a rate of—1.49 hPa/100 a since 1871.The method of wavelet analysis wasapplied to detect the low frequency characteristics of AOIs,and it is clear that there are decadalvariations in the time scale of about 20—30 years in all 4 indices.Since the averaged value of 4months could explain most variance of annual AOI.thus it is reasonable using the mean AOI ofJanuary,April,July and October to represent the annual AOI.The value of annual AOI was muchlower in periods of about 1894—1901.1910—1935,and was much higher in periods of about 1880—1893,1936—1945.The positive anomalies were remarkable since 1980s.Power spectra ofmodeled AOIs from Hadley Center Control Integration(HCCI)were compared with theobservations,it is found that the interannual time scale frequencies were more significant thandecadal frequencies.But.there also had the possibility for generating weak deeadal fluctuations insome special months and years by Hasselmann mechanism. This paper summarizes the long term and decadal time scale fluctuations of atmospheric circulation in the Southern (Hemisphere) middle and high latitudes.The long time series of Antarctic Oscillation Indices (AOIs) were established for January.April.July and October basedon the historical sea level pressure maps and NCEP / NCAR reanalysis surface pressure data set. Itwas found that the AOI of January had upward trend at a rate of 1.17 hPa / 100 a.and July haddownward trend at a rate of-1.49 hPa / 100 a since 1871.The method of wavelet analysis wasapplied to detect the low frequency characteristics of AOIs, and it is clear that there are decadalvariations in the time scale of about 20-30 years in all 4 indices. Since the averaged value of 4months could not explain most variance of annual AOI. Thus it is reasonable to use the mean AOI of January, April, July and October to represent the the AOI. The value of the annual AOI was muchlower in periods of about 1894-1901. 1910-1935, and was much higher in period s of about 1880-1893, 1936-1945. The positive anomalies were phenomenal since 1980s. Power spectra of modeled AOIs from Hadley Center Control Integration (HCCI) were compared with theobservations, it is found that the interannual time scale frequencies were more significant thandecadal frequencies .But.there also had the possibility for generating weak deeadal fluctuations insome special months and years by Hasselmann mechanism.