The traditional iterative learning control (ILC) algorithm improves the control performance by updating the control input to implicitly compensate the periodic uncertainties. In order to enhance the convergence rate of ILC, a new concept, iterative extended state observer (IESO), is presented which can estimate explicitly the periodic uncertainties during the process of iterations and be used to update the control input directly. The explicit estimation of the uncertainty by the linear IESO in the iteration domain is used to construct a new ILC algorithm based on active disturbance rejection (ADR). The ADR-based ILC algorithm and its corresponding theorem are given in detail and proven by using Lyapunov-like approach. Simulation results verify the effectiveness of the proposed ILC algorithm, and the iterative learning efficiency is improved greatly by using ADR-based ILC algorithm.