The concept of hierarchical control is employed to develop a two-level control model of human performance in complex tracking tasks. The lower level part of the model is composed of parallel suboptimal controllers describing, respectively, the human performance in subsystems, whereas the upper level is a simple rule-based compensator designed for modeling human behavior in coordinating the subsystems. The model has been used to study human performance in a manned anti-aircraft artillery tracking system. Model predictions of tracking errors and some other system variables for several target flyby and maneuvering trajectories are shown to be in good agreement With the empirical data obtained from experiments. It is concluded that the human model presented in this paper can be used with confidence in predictions of man-machine tracking system performance for system analyses and evaluations.