OBJECTIVE Studies have already investigated vestibulo-ocular reflex (VOR) responses in elderly subjects, mostly at low frequencies (<1 Hz) during passive head turns, or continuous active head turns in a rotational chair. However, natural head movements usually occur at frequencies above 1 Hz and at varying rates, rather than at continuously increasing rates as tested in most studies to date. The aim of this study was to compare VOR responses within or between normal and bilateral peripheral vestibular hypofunction (BPVH) elderly subjects with a computer base program incorporating random active high-frequency head movements. METHODS Seventeen senior patients with bilateral peripheral vestibular hypofunction and 13 age-matched paid healthy subjects participated in this study. METHODS All the subjects performed 8s of active, side-to-side head motions at 1 Hz, 2 Hz and 3 Hz in response to four experimental conditions: (A) a stationary visual target; (B) a stationary visual target and a moving target with a horizontal trajectory; (C) a stationary visual target and a moving target with a spiral trajectory; and (D) a stationary visual target with combined horizontal- and vertical-spiral moving targets. RESULTS Repeated-measures ANOVA showed that across the different frequencies of headshaking, a significant increase took place in the phase lag of the BPVH subjects (p<0.001) compared with the normal controls. Post hoc analysis showed significant within-group differences in BPVH subjects (1 Hz vs 2 Hz, 2 Hz vs 3 Hz) and normal controls (1 Hz vs 2 Hz, 1 Hz vs 3 Hz). A significant VOR gain took place among the BPVH subjects at 3 Hz VAT (p<0.001) as compared with 1 Hz VAT and no significant differences in VOR gain could be detected among the controls at either of these frequencies. Repeated-measures ANOVA revealed that under our test conditions, significant main effect was noted for VOR asymmetry (F=4.45, p<0.05). Post hoc analysis showed significant within-group differences in the BPVH subjects (test A vs test B, test A vs test C, and test A vs test D). CONCLUSIONS Our results showed that the VAT paradigm can be improved by using concurrent horizontal and vertical moving targets. The VOR phase may be useful for differentiating VAT responses between BPVH and healthy elderly subjects. Moreover, the results of this study demonstrate that gains in VOR at different frequencies of headshaking and asymmetry during different test conditions can be useful parameters for within-group assessment.