Abstract: This paper presents a hybrid intelligent model for multi-stand cold tandem rolling dynamic process, a system with compositive complexity including nonlinearity, multi variable, strong coupling, varying parameter as operation condition variation. The proposed model consists of the based state-space model based on linearized physical equations near rolling operating points, the exit strip speed RBF models and estimation approach for time lag of entry thickness, case-based reasoning technique based multi models selection algorithm for different operating points. With the actual thickness system setting and plate disturbance, the simulation results of roll force, strip thickness and tension have the same dynamic characteristics with theactual process data, and the error of each variable is less than 20%.