Abstract: Based on daily rainfall data over East Asia for the period of 1971 to 2007, nonlinear correlations between different grid points are calculated by event synchronization method, and an extreme rainfall network over East Asia is built. By using complex network method, regional characteristics of extreme rainfall over East Asia are analyzed and a dynamics prediction model is constructed. Spatial distribution of degree shows that spatial synchronization is better in northern inland region than in coastal region. It is found that grid points in different regions have different spatial correlation extents, and the northern inland has larger spatial correlation extent than the coastal region. Furthermore, the results show that the prediction accuracy is higher in coastal region than in northern inland region. This is partly because of the short average link distance among grid points, and partly because of the extreme precipitation intensity. The prediction model constructed from the perspective of space and time continuity has the ability to predict extreme rainfall in East Asia on a certain level, and it has some potential application values in the research of extreme rainfall.