This observational study aimed at investigating pain in stroke patients with upper limb spastic dystonia. Forty-one consecutive patients were enrolled. A 0–10 numeric rating scale was used to evaluate pain at rest and during muscle tone assessment. Patients were asked to indicate the most painful joint at passive mobilization (shoulder, elbow, wrist-fingers). The DN4 questionnaire was administered to disclose neuropathic pain. All patients were assessed just before and 1 month after incobotulinumtoxin-A treatment. Pain was present in 22 patients, worsened or triggered by passive muscle stretching. DN4 scored < 4 in 20 patients. The most painful joints were wrist–fingers in 12 patients, elbow in 5 patients and shoulder in the remaining 5 patients. Both elbow and wrist–fingers pain correlated with muscle tone. BoNT-A treatment reduced pain in all the joints, including the shoulder. We discussed that nociceptive pain is present in a vast proportion of patients with upper limb spastic dystonia. BoNT-A treatment reduced both spastic dystonia and pain in all the joints but the shoulder, where the effect on pain could be mediated by the reduction of pathological postures involving the other joints.