The present study aimed to investigate the relationship between neuropathic symptoms (i.e. pain and paraesthesia/dysaesthesia) and structural damage and functional alterations of spinal sensory tracts in patients with syringomyelia. Three-dimensional fibre tracking of the cervical spinal cord (at level C3-C4), electrophysiological assessments of nociceptive (laser-evoked potentials) and non-nociceptive (somatosensory-evoked potentials) pathways and quantitative sensory testing were carried out in 37 patients with syringomyelia, 27 with neuropathic pain and 21 controls. Four regions of the body (both hands and shoulders) were systematically examined with laser-evoked potentials and quantitative sensory testing, and somatosensory-evoked potentials were induced from both hands. The diffusion tensor imaging variables investigated included the mean fractional anisotropy, the mean apparent diffusion coefficient and the number of reconstructed nerve fibres of the tracts located within three volumes of interest (full spinal section, anterior cord and posterior cord). Consistent with the results of previous studies, patients with or without neuropathic pain were indistinguishable on the basis of quantitative sensory testing, laser-evoked and somatosensory-evoked potentials and three-dimensional fibre tracking analyses. However, in patients with neuropathic pain, higher average daily pain intensity was correlated with greater structural damage to the spinal cord, as assessed by fractional anisotropy (Spearman's ρ = -0.64, P = 0.020) and the number of reconstructed nerve fibres (r = -0.75; P = 0.020) of the full spinal cord. The number of reconstructed nerve fibres was negatively correlated with two neuropathic dimensions, i.e. 'deep spontaneous pain' (r = -0.59, P = 0.040) and 'paraesthesia/dysaesthesia' (i.e. pins and needles/tingling) (r = -0.67, P = 0.020), suggesting that various pain descriptors have distinct underlying mechanisms. Patients with both spontaneous and evoked pain clearly differed from patients with spontaneous pain only. Patients with spontaneous pain only had more severe spinal cord damage, and the correlation between average daily pain intensity and fractional anisotropy of the full spinal cord was particularly strong in this subgroup of patients (Spearman's ρ = -0.93, P = 0.008). By contrast, patients with both spontaneous and evoked pain had not only less structural spinal cord damage, but also better preserved spinothalamic and lemniscal tracts on quantitative sensory testing and electrophysiological testing. These data showed, for the first time, a direct relationship between central neuropathic pain and objective markers of spinal cord damage, and confirmed the clinical relevance of 3D fibre tracking for the sensory assessment of patients with a spinal cord lesion.