This paper examines the topography of neuronal degeneration in the central nervous system of the dystonia musculorum (dt) mutant mouse, revealed by selective silver impregnation, specific histochemical staining and electron microscopy. Neuronal lesions have been observed exclusively in the spinal cord, the medulla and the anterior lobe of the vermis. In the spinal cord, axonal degeneration was maximal among large and medium-sized primary sensory fibers, whereas thin caliber primary afferents were unaffected, with the exception of those containing acid phosphatase activity. In regions of laminae VI to VIII that receive numerous degenerative primary afferents, neurons undergoing different phases of degeneration (chromatolysis, lipid accumulation, dark shrunken necrosis) were constantly found. Most of the latter belonged to spinocerebellar neurons, owing to the presence of fiber degeneration in both spinocerebellar tracts and mossy fiber degeneration in the anterior vermal lobe. In the medulla only axonal degeneration was observed and was confined to three fiber systems: the dorsal column pathway, the sensory trigeminal fibers (both from the trigeminal ganglion and from the mesencephalic trigeminal nucleus), and the spinocerebellar fibers entering the cerebellum through the inferior and superior cerebellar peduncles. This study also suggests a simple pathophysiological mechanism for the onset and the progression of the degeneration: dystonic gene action would affect perinatally specific classes of sensory receptors, producing the degeneration of the nerve terminals and, progressively, the cell death of the sensory ganglion cells at their origin. This retrograde death, which results in the massive and early deafferentation of spinocerebellar neurons, would provoke, trans-neuronally, the impairment of these second order sensory neurons and the progressive degeneration of the spinocerebellar system. The close resemblance of the neuropathology of the mutant mouse to Friedreich's ataxia (the commonest form of human degenerative ataxic disorders) allows one to suppose that the dystonic mouse may be an optimal animal model for studying the genetic basis and the pathophysiological mechanisms of this form of human ataxia.