In view of the strong coupling, multivariable and other nonlinear factors of the rolling mill hydraulic position closed-loop system, the accurate modeling is difficult and there is no self-renewal learning ability,the model-free adaptive iterative learning control (MFAILC) was applied to the rolling mill hydraulic position closed-loop system. Since the error convergence time of the MFAILC algorithm is long, the high-order pseudo-partial derivative estimation algorithm was used to improve the convergence speed of the system. At the same time, aiming at the problems of poor anti-interference and easy to cause control deviation in the control process of MFAILC algorithm, combining with the advantages of strong robustness and simple structure of internal model control, it was introduced into MFAILC to improve the control structure of the algorithm. The simulation results show that the improved algorithm has faster convergence speed and higher control accuracy, the anti-interference performance of the system is also enhanced.