Background. Three-dimensional motion analysis represents a quantitative and objective approach to assess spatio-temporal and kinematic changes in Parkinson s disease (PD). However, these parameters, focusing on a specific body segment, provide only segmental information, discarding the complex whole-body patterns underlying the motor impairment. We aimed to assess how levodopa intake affects the whole body, large-scale kinematic network in PD. Methods. Borrowing from network theory, we used the kinectome framework by calculating the Pearson s correlation coefficients between the time series acceleration of 21 bone markers. Then, we performed a topological analysis to evaluate the large-scale interactions between body elements. Finally, we performed a multilinear regression analysis in order to verify whether the kinectome s topological features could predict the clinical variation before and after levodopa intake. Results. PD patients showed lower nodal strength (i.e., lower synchronization) in the upper body in the medio-lateral acceleration while in on-state with respect to the off state (p-head=0.048; p-C7=0.032; p-T10=0.006). On the contrary, PD patients in on state displayed higher nodal strength (i.e., higher synchronization) of both elbows (right, p=0.002; left, p=0.005), wrists (right, p=0.003; left, p=0.002) and knees (right, p=0.003; left, p=0.039) in the antero-posterior acceleration. Furthermore, the predictive analysis revealed that the nodal strength variations of the arms, following levodopa intake, significantly predicted the clinical variations assessed through the UPDRS-III (R2=0.65; p=0.025). Conclusions. PD patients in the on-state showed less rigidity during walking, proportional to the UPDRS-III variation. More importantly, we showed that levodopa induces an improvement of the whole body, large-scale kinematic pattern.