The Southern Oscillation (SO) is characterized by a temporal signal that dominates the variation of sea surface temperature (SST), pressure, and other fields in “core regions,” which are mostly in or near the equatorial Pacific. It involves persistence, high interannual variability, and high correlations between fields. All these characteristics vary with season, being most marked around November and weakest around April. These phenomena are best explained in terms of a positive feedback relationship between the equatorial east Pacific SST and the Walker circulation, in which the feedback varies with season. The relationship between SST anomalies and cloudiness varies with season in a sense that could account for the required variation in feedback. The SO involves simultaneous teleconnections that can probably be explained by atmospheric dynamical processes. There are also lag teleconnections that call for different explanations. For example, tropical Indian Ocean SST tends to be low several months after high SO index. The explanation for this could involve the sequence: high index → low pressure over Indian Ocean → increased convergence → increased cloudiness → reduced net radiation at sea surface → lower SST. Of particular interest are lag relationships in which some factor is correlated better with later SO than with simultaneous SO. There is evidence of such factors in the southeast Pacific, North Australia, and east equatorial Atlantic regions in certain seasons. Plausible physical hypotheses are available to account for the relation-ships. These factors represent predictors of the SO, and deserve detailed study both to improve seasonal forecasts and to shed light on the mechanisms of SO fluctuations. Some relationship seem to involve lag correlations in both senses, and thereby imply negative feedback relationships that could be the cause of the tendency for the SO to fluctuate.