An ultra-precision stage system is a critical component of a wafer scanner, including the wafer stage and the reticle stage. The synchronous performance of the wafer stage and reticle stage significantly affects the overlay and critical dimension uniformity of the machine. A moving-average-and-standard-deviation-based iterative learning control (MASD-ILC) method is proposed, which can reduce synchronous errors and suppress thrust ripple. The convergence principle and conditions of the method along the time axis and the iteration axis are proved. The effects of the learning gain and weights is discussed. The effectiveness of the proposed method is verified through simulations. Compared with the e-ILC, the MASD-ILC demonstrates faster convergence speed, smaller convergence error, and better robustness. The MASD of the system based on the MASD-ILC is reduced from 31.56 nm to 0.10 nm. Considering the thrust perturbation and model uncertainty, the convergence speed and convergence error of the system based on the MASD-ILC remain unaffected.