Squirrel monkeys were trained using newly developed visual-vestibular mismatch paradigms to test the asymmetrical simultaneous induction of vertical vestibuloocular reflex (VOR) gain changes in opposite directions (high and low) either in the upward and downward directions or in response to high- and low-frequency stimuli. The first paradigm consists of sinusoidal head movement [A sin(omegat)] and a full rectified sinusoidal optokinetic stimulus [+/-|A sin(omegat)|], whereas the second paradigm consists of the sum of two sinusoids with different frequencies [A sin(omega(1)t) + A sin(omega(2)t) for head motion and +/-[A sin(omega(1)t) - A sin(omega(2)t)] for the optokinetic stimulus, omega(1) = 0.1pi, omega(2) = 5pi]. The first paradigm induced a half rectified sinusoidal eye-velocity trace, i.e., suppression of the VOR during upward head motion and enhancement during downward head motion or vise versa, whereas the second paradigm induced suppression of the VOR at the low-frequency omega(1) and enhancement at the high-frequency omega(2) or vise versa. After 4 h of exposure to these paradigms, VOR gains of up and down or high and low frequency were modified in opposite directions. We conclude that the monkey vertical VOR system is capable of up-down directionally differential adaptation as well as high-low frequency differential adaptation. However, experiments also suggest that these gain controls are not completely independent because the magnitudes of the gain changes during simultaneous asymmetrical training were less than those achieved by symmetrical training or training in only one of the two components, indicating an influence of the gain controls on each other. These results confine the adaptive site(s) responsible for vertical VOR motor learning to those that can process up and downward or low- and high-frequency head signal separately but not completely independently.