Summary Background Severe forms of primary dystonia are difficult to manage medically. We assessed the safety and efficacy of pallidal neurostimulation in patients with primary generalised or segmental dystonia prospectively followed up for 5 years in a controlled multicentre trial. Methods In the parent trial, 40 patients were randomly assigned to either sham neurostimulation or neurostimulation of the internal globus pallidus for a period of 3 months and thereafter all patients completed 6 months of active neurostimulation. 38 patients agreed to be followed up annually after the activation of neurostimulation, including assessments of dystonia severity, pain, disability, and quality of life. The primary endpoint of the 5-year follow-up study extension was the change in dystonia severity at 3 years and 5 years as assessed by open-label ratings of the Burke–Fahn–Marsden dystonia rating scale (BFMDRS) motor score compared with the preoperative baseline and the 6-month visit. The primary endpoint was analysed on an intention-to-treat basis. The original trial is registered with ClinicalTrials.gov (NCT00142259). Findings An intention-to-treat analysis including all patients from the parent trial showed significant improvements in dystonia severity at 3 years and 5 years compared with baseline, which corresponded to −20·8 points (SD 17·1; −47·9%; n=40) at 6 months; −26·5 points (19·7; −61·1%; n=31) at 3 years; and −25·1 points (21·3; −57·8%; n=32). The improvement from 6 months to 3 years (–5·7 points [SD 8·4]; −34%) was significant and sustained at the 5-year follow-up (–4·3 [10·4]). 49 new adverse events occurred between 6 months and 5 years. Dysarthria and transient worsening of dystonia were the most common non-serious adverse events. 21 adverse events were rated serious and were almost exclusively device related. One patient attempted suicide shortly after the 6-month visit during a depressive episode. All serious adverse events resolved without permanent sequelae. Interpretation 3 years and 5 years after surgery, pallidal neurostimulation continues to be an effective and relatively safe treatment option for patients with severe idiopathic dystonia. This long-term observation provides further evidence in favour of pallidal neurostimulation as a first-line treatment for patients with medically intractable, segmental, or generalised dystonia. Funding Medtronic.