Patients with dystonia display a number of disturbances in the cognitive processing of movements, such as movement simulation and prediction, but whether these deficits point to a deeper rooted disturbance of perceptual bodily representations remains unknown. A useful way to investigate the sense of body ownership is the rubber hand paradigm, in which an illusion of ownership is established by synchronous stroking of the participants’ real unseen hand and a visible fake hand, whereas similar asynchronous stroking does not bring about the illusion. This paradigm allows testing of both the subjective experience of feeling ownership over the rubber hand and the proprioceptive relocation of the real unseen hand towards the viewed rubber hand. Previous studies have mapped these different aspects onto two anatomically distinct neuronal substrates, with the ventral premotor cortex processing the illusory feeling of ownership and the inferior parietal lobule and cerebellum processing proprioceptive drift. We applied the rubber hand illusion task to healthy subjects and to patients affected by two different types of focal dystonia—one specifically affecting the hand (focal hand dystonia) and one not affecting the hand (torticollis and blepharospasm). Results showed that in patients with focal hand dystonia, the proprioceptive drift was selectively disrupted on the dystonic hand while the subjective experience of the illusion was retained. In the non-dystonic hand and in the other two groups (non-hand dystonia and healthy subjects), the rubber hand illusion resembled the typical pattern with synchronous stroking eliciting the illusion. These findings provide support for the contention that the mechanisms underlying the presence of the illusory feeling of ownership and the proprioceptive drift are different. Selective impairment of the limb recalibration on the dystonic hand points to underlying deficits in integrating the visual-tactile input with the proprioceptive information in order to update the current body position and may support a model linking dystonia to dysfunctions in a network comprising the inferior parietal cortex and the cerebellum.