The reproducibility of the Southern Hemisphere Annular Mode (SAM) is examined by using the NCAR CAM2 model forced with historical sea surface temperature covering the time period of 1950--2000. There is some correspondence (low however significant correlation) between the simulation and observation in the temporal evolution of the SAM, indicating if global SSTs are known and prescribed to the model, there would be some predictability to the SAM. The reproducibility is greater during austral summer than winter, and the source of much of the reproducibility is tropical Pacific SST, with equatorial Pacific warm event corresponds to a negative phase SAM in austral summer. There is a hint that this result is model independent. Working with ensembles improves the reproducibility.