Abstract: The semiannual oscillation (SAO) of the atmospheric circulations in the extratropic regions is one of the unique characteristics in the Southern Hemisphere. Based on the NCEP/NCAR reanalysis datasets of the sea level pressure (SLP) and the geopotential height from January 1979 to December 2000 the oscillation is studied. Results show that the SAO is most active in the extratropic regions in the lower troposphere but the tropics in the middle-higher level. The extratropical SAO of SLP is found to be most significant in 40°S and 65°S, i.e., the two branches of the Antarctic oscillation (AAO) or named as the Southern Hemisphere annular mode (SAM), where each variance percentage of the half-year period exceeds 70%. That is to say that the anti-phase relationship for the annual cycle of SLP between the two latitudes is primarily caused by their SAO components. But the SAO is not remarkable between these two latitudes, especially in 55°S, i.e., the node position of the standing wave of the AAO. In 55°S, the variance percentage of SAO is less than 20% while that of the annual cycle is greater than 70%. Results also indicate that the phases at the latitudes of 40°S and 65°S are inconsistent, with maximam variance percentages occurring in March and September while minimums in June and December respectively in 40°S, but maximums in January and July and minimums in April and October respectively in 65°S. In other words, the phase of the SAO in the near-polar latitude has a one-month lag than in the middle counterpart.At present, most of the general circulation models (GCMs) are poor in simulating the characteristics of the extratropical SAO in the Southern Hemisphere. So the SAO is used widely as a criterion to verify a GCM's simulation skill. Based on the above results, a nine-level atmospheric general circulation model developed in the Institute of Atmospheric Physics (IAP 9L AGCM) is employed to simulate the SAO. In general, the model is successful in simulating the SAO of SLP along 65°S, with its intensity some what weaker than the observation. On the other hand, the IAP 9L AGCM fails to simulate the SAO along 40°S.