This paper presents the results of applying the Iterative Learning Control algorithms to a Twin‐Rotor Multiple‐Input Multiple‐Output System (TRMS) in order to achieve high performance in repetitive tracking of trajectories. The plant, which is similar to a prototype of helicopter, is characterized by its highly nonlinear and cross‐coupled dynamics. In the first phase, the system is modelled using the Lagrangian approach and combining theoretical and experimental results. Thereafter, a hierarchical control architecture which combines a baseline feedback controller with an Iterative Learning Control algorithm is developed. Finally, the responses of the real device and a complete analysis of the learning behaviour are exposed.