This article first investigates robust iterative learning control (ILC) problem of a class of large-scale interconnected systems, which consist of many subsystems described by two-dimensional linear discrete first Fornasini-Marchesini systems with iteration-dependent uncertainties arising from not only boundary states, disturbances, but also reference trajectory. A decentralized P-type ILC law without any information exchanges with other subsystems is proposed such that the ultimate ILC tracking error of each subsystem can converge to a bounded range, the bound of which depends continuously on the bounds of all iteration-dependent uncertainties considered. Especially, if these iteration-dependent uncertainties are convergent progressively along the iteration direction, perfect ILC tracking on 2-D reference trajectory can be obtained.Additionally, a modified ILC law with compensation technique is used to a class of large-scale interconnected systems composed of many subsystems represented by two-dimensional linear discrete second Fornasini-Marchesini systems. Two simulation examples are used to demonstrate the effectiveness and validity of the obtained ILC results. Finally, some comparative discussions are given.