The Vestibular Autorotation Test (VAT) is a relatively easy method for testing the Vestibulo Ocular Reflex (VOR) at higher frequencies by asking the patient to actively shake his/her head from side to side (jaw) with increasing frequency. In this study we investigated reproducibility and clinical applicability of this method and quantified gain and phase of the VOR in normal subjects. We studied the VOR in 33 subjects over a frequency range from 2 to 6 Hz. Ten subjects were investigated with an angular rate sensor mounted on a headband. In this group the velocity gain tended to decrease at higher frequencies and phase lag increased slightly. In a second experiment (2 subjects) head velocity was detected simultaneously by an angular sensor fixed to a headband and an identical sensor attached to a bite board. It was shown that the gain of the VOR in the first group was underrated and phase shift overrated due to slippage of the headband fixed rate sensor. Therefore in a third experiment, 21 subjects were investigated with the rate sensor attached to a bite board only. In this group VOR gain was close to unity and phase was close to 0 degrees (180 degrees out of phase) from 2 up to 6 Hz. In contrast to other studies, a considerable number of subjects were unable to reach shaking frequencies above 4 Hz. Our results agree with those of recently published reports and disagree with others, both using a headband. This is interpreted as being due to difficulties in sensing head velocity accurately. CONCLUSIONS vestibular autorotation is an accurate and reproducible way of testing the high frequency range of the vestibulo-ocular reflex, provided that head velocity is measured by a bite board. Our present work is focused on determining the high frequency characteristics of the VOR in patients suffering from vertigo.