In order to study the predictability of ENSO and its spatial and temporal changes, the mutual information and Cao methods, which can ascertain the parameters in the phasespace reconstruction, are used to analyze the time series of sea surface temperature anomaly (SSTA). The results of the mutual information method show that the calculated delay time of the longer time series is more confident with the theoretical value. The results of the Cao method reveal that the randomicity of the SSTA series is stronger than that of the sea clutters and the certainty of the SSTA series in the eastern equatorial Pacific is stronger than that in the western Pacific.