The perception of body verticality (subjective postural vertical, SPV) was assessed in normal subjects and in patients with peripheral and central vestibular lesions and the data were compared with conventional neuro-otological assessments. Subjects were seated with eyes closed in a motorized gimbal which executed cycles of tilt at low constant speed (1.5° s−1), both in the frontal (roll) and sagittal (pitch) planes. Subjects indicated with a joystick when they entered and left verticality, thus defining a sector of subjective uprightness in each plane. The mean angle of tilt (identifying a bias of the SPV) and the width of the sector (defining sensitivity of the SPV) were then determined. In normal subjects, the angle of the ‘verticality’ sector was 5.9° for pitch and roll. Patients with bilateral absence of vestibular function, patients with vertigo, i.e. acute unilateral lesions, benign paroxysmal positional vertigo (BPPV) and Ménière's disease, and patients with positionally modulated up-/ downbeat nystagmus all had enlarged sectors (i.e. loss in sensitivity). Mean sector angle in these groups ranged from 7.8 to 11° and the abnormality was present both in pitch and roll, regardless of the direction of nystagmus or body sway. Patients with chronic unilateral peripheral vestibular lesions and those with position-independent vertical nystagmus had normal sensitivities. No significant bias of the SPV was found in any patient group, not even those with acute unilateral vestibular lesions who had marked tilts of the subjective visual vertical (SVV). Complementary experiments in normal subjects tested under galvanic vestibular or roll-plane optokinetic stimulation also failed to show biases of the SPV. In contrast, a significant bias in the SPV could be induced in normal subjects by asymmetric cycles of gimbals tilt, presumably by proprioceptive adaptation. The following conclusions can be drawn. (i) The perception of body verticality whilst seated is mainly dependent on proprioceptive/contact cues but these are susceptible to tilt-mediated adaptation. (ii) Vestibular input improves the sensitivity of the SPV, even in vestibular disorders, as long as the abnormality is stable. (iii) There can be marked dissociation between vestibulo-motor (ocular and postural) phenomena and the perception of body verticality, and between the SPV and SVV. (iv) The postural sway asymmetries in patients with peripheral and central vestibular lesions, like those induced by galvanic or optokinetic stimulation in normal subjects, are not consequences of changes of the SPV.