A novel combination of PI control and iterative learning control (ILC), referred to enhanced ILC-based PI control, was proposed in this study. This algorithm could be used on a class of multi-input multi-output batch processes with modeling uncertainties and unknown disturbances. Based on a two-dimensional (2D) Roesser's system description, a sufficient condition for robust asymptotical stability of the closed-loop system was derived in this paper. Furthermore, the monotonic convergence of the closed-loop system in the batch direction has been proved. Applications on a three-tank system show that the proposed method can achieve the design objective well, with performance improvement along both time and batch direction, and also owns good robustness to uncertainties and non-repetitive disturbances. (C) 2012 Elsevier Ltd. All rights reserved.