Background: Subthalamic nucleus (STN) deep brain stimulation (DBS) improves the cardinal symptoms of Parkinson’s disease (PD). However, the therapeutic mechanisms are incompletely understood. By leveraging patient-specific brain responses to DBS using functional magnetic resonance imaging (fMRI) acquired during stimulation, we identify and validate symptom-specific networks associated with clinical improvement. Methods: Forty PD patients with STN-DBS were enrolled for fMRI using a 30-sec DBS-ON/OFF cycling paradigm. The four cardinal motor outcomes of PD were chosen a priori and measured using the Movement Disorder Society-Sponsored Revision of the Unified Parkinson’s Disease Rating Scale, part III (MDS-UPDRSIII): axial instability, tremor, rigidity, bradykinesia. Stimulation-dependent changes in blood oxygen level-dependent (BOLD) signal were correlated with each symptom. Results: The relationship between BOLD response and outcomes revealed significant networks of clinical response (p<0.001). Using BOLD responses from the network hubs, each symptom-specific model was significantly predictive of actual improvement: axial instability (R2=0.38, p=0.000026), bradykinesia (R2=0.29, p=0.00033), rigidity (R2=0.40, p=0.000013), tremor (R2=0.26, p=0.00073). Conclusions: Using patient-specific imaging, we provide evidence of an association between DBS-evoked fMRI response and individual symptom improvement. Brain networks associated with clinical improvement were different depending on the PD symptom examined, suggesting the presence of symptom-specific networks of efficacy which may allow personalization of DBS therapy.