We describe the clinical features of 103 patients presenting with fixed dystonia and report the prospective assessment and investigation of 41 of them. Most patients were female (84%) and had a young age of onset [mean 29.7 (SD 13.1) years]. A peripheral injury preceded onset in 63% and spread of dystonia to other body regions occurred in 56%. After an average follow-up of 3.3 years (overall disease duration 8.6 years), partial (19%) or complete (8%) remission had occurred in a minority of patients. The fixed postures affected predominantly the limbs (90%), and rarely the neck/shoulder region (6%) or jaw (4%). In the prospectively studied group, pain was present in most patients and was a major complaint in 41%. Twenty percent of patients fulfilled criteria for Complex Regional Pain Syndrome (CRPS). No consistent investigational abnormalities were found and no patient tested (n = 25) had a mutation in the DYT1 gene. Thirty-seven percent of patients fulfilled classification criteria for documented or clinically established psychogenic dystonia; 29% fulfilled DSM-IV (Diagnostic and statistical manual of mental disorders, 4th edition) criteria for somatization disorder, which was diagnosed only after examination of the primary care records in many cases; and 24% fulfilled both sets of criteria. Ten percent of the prospectively studied and 45% of the retrospectively studied patients did not have any evidence of psychogenic dystonia, and detailed investigation failed to reveal an alternative explanation for their clinical presentation. Detailed, semi-structured neuropsychiatric assessments in a subgroup of 26 patients with fixed dystonia and in a control group of 20 patients with classical dystonia revealed dissociative (42 versus 0%, P = 0.001) and affective disorders (85 versus 50%, P = 0.01) significantly more commonly in the fixed dystonia group. Medical and surgical treatment was largely unsuccessful. However, seven patients who underwent multidisciplinary treatment, including physiotherapy and psychotherapy, experienced partial or complete remission. We conclude that fixed dystonia usually, but not always, occurs after a peripheral injury and overlaps with CRPS. Investigations are typically normal, but many patients fulfil strict criteria for a somatoform disorder/psychogenic dystonia. In a proportion of patients, however, no conclusive features of somatoform disorder or psychogenic disorder can be found and, in these patients, whether this disorder is primarily neurological or psychiatric remains an open question. Whilst the prognosis is overall poor, remissions do occur, particularly in those patients who are willing and able to undergo multidisciplinary treatment including physiotherapy and psychotherapy, suggesting that this type of treatment should be recommended to these patients.