Although facilitation of the cortico-spinal system during action observation is widely accepted, it remains controversial whether this facilitation reflects a replica of the observed movements or the goal of the observed motor acts. In the present transcranial magnetic stimulation study, we recorded motor evoked potentials from two hand muscles (first dorsal interosseous and abductor digiti minimi) while 22 healthy participants observed a hand reaching towards and grasping a bottle. To test for alternative coding levels (goal vs. movement), three relevant aspects were systematically manipulated: the type of observed movement (precision grip or whole hand grasping), situational context (bottle positioned in front of or behind a wall-like barrier), and processing stage (transcranial magnetic stimulation pulse delivered at the onset of the movement or at the moment of contact between the fingers and the object). At movement onset, motor evoked potential responses reflected the program necessary to achieve the action goal within the situational context. During movement observation, however, the type of observed movement was taken into account and a transition towards a movement-related modulation was observed. These results suggest that, rather than being exclusive alternatives, goal coding and movement coding may relate to different processing stages.