We have carried out a prospective study of selective peripheral denervation (SPD) in cervical dystonia (CD) patients with primary or secondary botulinum toxin (BT) treatment failure using independent standardized assessment. Patients referred for surgery had a standardized clinical examination, neck muscle EMG, videofluoroscopic swallow and CT of the cervical spine, and were selected for surgery on the basis of the results of these investigations. CD severity, disability and pain were assessed preoperatively and at 3, 6, 9, 12 and 18 months postoperatively using the Toronto Western Spasmodic Torticollis Rating Scale (TWSTRS). Severity of head tremor and dysphagia were scored using established rating scales. Additionally, psychosocial function was assessed in a representative subsample of patients (n = 12) using several established questionnaires. Of the 62 patients who were assessed, 22 (35.5%) were not offered surgery, most commonly because of widespread dystonia. Of the remaining 40 patients, 37 have so far had surgery, 31 of whom have been followed up for at least 1 year, and 15 for 18 months after surgery (mean follow-up duration 16.7 months). Using the TWSTRS global outcome score, 68% of patients derived functionally relevant improvement at 12 months after surgery. In the entire operated group, total TWSTRS scores were reduced by 30% at 6 and 12 months after surgery (P < 0.0001). The subscores for severity, disability and pain were reduced by 20, 30 and 40%, respectively, at 6 months (P < or = 0.01) and 20, 40 and 30%, respectively, at 12 months (P < 0.01). Pain increased over time, which appeared to result from muscle reinnervation. TWSTRS scores were not significantly improved in the six patients with primary BT treatment failure. Head tremor did not change. There was a significant improvement of body concept, perceived disfigurement, stigma, and quality of life in the 12 patients whose psychosocial function was assessed. Preoperative disability and restriction of head movement were negatively correlated and the initial response to BT treatment positively correlated with global outcome score. Spread or deterioration of dystonia elsewhere in the body occurred in three patients, with unpleasant sensory symptoms in denervated posterior cervical segments occurring in 14. Ten patients developed mild to moderate dysphagia, and two developed severe dysphagia. We conclude that SPD is an effective treatment for patients with secondary, but probably not for those with primary, BT treatment failure. Reinnervation is not infrequent and can compromise outcome. Postoperative morbidity is low, but there is a risk of dysphagia.