A growing number of studies postulate the use of music to improve motor control in patients with Parkinson's disease (PD). The effects of music are greatly variable from one individual to the other and do not always reach the expected benefits. This study aimed to optimize the use of music in the management of movement disorders inherent to PD in a handwriting task. We developed and tested musical sonification (MS), a method that transforms in real-time kinematic variables into music. Twelve patients with PD, on medication, and 12 healthy controls were recruited in a pretest/training/posttest design experiment. Three training sessions were compared, for which participants were asked to produce graphomotor exercises: one session with music (unrelated to handwriting), one with MS (controlled by handwriting), and one in silence. Results showed that the performance in training was better under MS than under silence or background music, for both groups. After training, the benefits of MS were still present for both groups, with a higher effect for PD patients than for control group. Our results provide a proof of concept to consider MS as a relevant auditory guidance strategy for movement rehabilitation in patients with PD.