Two experiments were conducted to examine the human ability to adapt to a perturbation in a synchronization task. Five experimental signal conditions were tested using random-dot kinematograms, representing four conditions with different coherence levels (100%, 50%, 30% and 10%) and one target-alone condition. Within one trial, increasing or decreasing the frequency of the sinusoidally moving signal dots abruptly in the midst of each trial provoked a perturbation. The first experiment was aimed to clarify the process of adaptation to the new frequency situation. The second experiment explored the role of visual feedback about the arm's position on the participants' ability to adapt after the perturbation had occurred. The results clearly demonstrated that the synchronization performance gradually declined in function of the increasing number of randomly moving noise dots. In the 50% coherence condition, the participants were not or only partially able to adjust their arm movements to the new frequency situation. In addition, the provision of enhanced visual feedback about the arm's failed to improve one's adaptive ability. In general, these findings provided evidence for the important role of perceptual constraints on perception-action coupling in this type of synchronization task.