The vestibulo-ocular reflex (VOR) serves to stabilize images on the retina. To maintain appropriate performance and minimize image slippage throughout life, the VOR is subject to long-term adaptive regulation in response to visual input. Adaptive changes in VOR gain (eye velocity/head velocity) can be evoked either by fitting subjects with magnifying, miniaturizing, or reversing spectacles during normal behavior or by moving a large visual field in or out of phase relative to the subject's head movement. These changes exhibit frequency-selectivity. Here, we examine the flexibility of VOR gains by causing VOR in similar directions to undergo different behavioral gain changes. Nine healthy adults, ranging in age from 24 to 38 (mean 28.5) with no history of neurotological symptoms participated in the study. All subjects demonstrated clinically normal functioning on a screening battery of tests that included combined neurologic and otologic physical examinations. Horizontal and vertical eye positions were recorded by bitemporal DC coupled electrooculography (EOG). The subject sat in a rotating chair. The axis of rotation of the body was always earth-vertical, the interaural axis crossing the axis of rotation of the chair. The head was positioned at 20 degrees down in all experiments and was stabilized in this position using a chin rest. The chair was 78 cm in diameter and was shielded by a half-cylindrical optokinetic screen positioned in front of the subjects. Random dot patterns were projected onto this screen. During per- and post-adaptation periods, goggles were fitted to ensure that the subject was in complete darkness and the chair was rotated sinusoidally. The amplitude of the rotating chair was 30 degrees and 60 degrees. Frequencies of rotation were 0.1 Hz, 0.2 Hz, 0.3 Hz and 0.4 Hz for amplitudes of 30 degrees and 0.1 Hz, 0.2 Hz, and 0.3 Hz for amplitudes of 60 degrees. To induce VOR adaptation, the retinal slippage velocity caused by the visual input of a large field was changed for short-term; the change was produced by a combination of sinusoidal head rotation and random dot patterns. During each adaptation session, the frequencies of sinusoidal head rotation were either 0.1 Hz or 0.3 Hz and the amplitude was 30 degrees. The random dot patterns were synchronized with the sinusoidal head rotation in the same direction to make the retinal slippage zero (x0 experiment) and in the opposite direction to double the retinal slippage (x2 experiment). Therefore, a total of four adaptation protocols were tested. The subjects were asked to fix their eyes on a single dot by looking straight ahead in the x0 experiment and to follow the dot in the random dot pattern in the x2 experiment. Each adaptation session lasted for 30 minutes. Each subject participated in couple of adaptation experiments everyday. The average VOR gain and phase lag were calculated using a Fourier transformation. Out of all the subjects who participated in the x2 adaptation experiment at 0.3 Hz with an amplitude of 30 degrees, seven subjects showed a steady increase in VOR gain during a couple of the trials. One out of the remaining two subjects showed a decrease in VOR gain in all three trials. Another subject showed an increase in VOR gain during three trials and a decrease in two trials. In the x2 adaptation experiment with a range of 30 degrees at 0.3 Hz (peak velocity: 28 degrees/s), the percent change in gain (post-pre/pre) was 133% at the same stimulation and 100% at 0.4 Hz (peak velocity: 37 degrees/s). The percent change in gain was 65% for amplitudes of 60 degrees at 0.1 Hz (peak velocity: 18 degrees/s) and 64% for amplitudes 60 degrees at 0.2 Hz (peak velocity: 37 degrees/s). In the x1 adaptation experiment (30 degrees at 0.3 Hz), the percent change in gain was -62% for the same conditions and -50% for amplitude of 60 degrees at 0.1 Hz and -30% for amplitudes of 60 degrees at 0.2 Hz. No change, in VOR gain was observed at the other frequencies. The