Abstract: Based on 1950-2000 monthaverage SST field from the NCEP/ NCAR reanalysis data, SST field series are spacetime decompounded using the EOF method. With the hypothesis of EOF space-mode being stable and unchangeable assumed and the dynamic system reconstruction idea introduced, the EOF time coefficient series is taken as the dynamical statistic model variable and the dynamical model parameters are optimized and retrieved to establish a nonlinear dynamic statistic model of EOF separated time coefficient series. By the model time integral and EOF time-space reconstruction, a mid-long-range forecast of SST field is performed. The test result shows that to the 1-6 month forecast, the SST field forecasted by our model are very similar to the actual SST field; to 7-15 month forecast, the basic conformation are approximately the same although the SST field forecasted by our model has some differences from the actual SST field; in particular, the configuration of the SST field more than 12 months can be accurately depicted. The forecast results for all the leading times can describe the characteristic of the El Nino in 1997 to some extent. The investigative method suggested a new approach to forecasting SST field and showed a better result. This nethodology and technique suggested in this paper for retrieving the parameters of the dynamical model of the SST field of the tropical Pacific Ocean El Nino /La Nina events show nice effect on the forecasting SST field of the tropical Pacific Ocean (especially in the medium and long term), offering the nice investigative search and method of the SST field of the tropical Pacific Ocean and the medium and long term forecast of the exceptional El Nino /La Nina events.