A recent study showed that the angular vestibuloocular reflex (VOR) can be better adaptively increased using an incremental retinal image velocity error signal compared with a conventional constant large velocity-gain demand (×2). This finding has important implications for vestibular rehabilitation that seeks to improve the VOR response after injury. However, a large portion of vestibular patients have unilateral vestibular hypofunction, and training that raises their VOR response during rotations to both the ipsilesional and contralesional side is not usually ideal. We sought to determine if the vestibular response to one side could selectively be increased without affecting the contralateral response. We tested nine subjects with normal vestibular function. Using the scleral search coil and head impulse techniques, we measured the active and passive VOR gain (eye velocity / head velocity) before and after unilateral incremental VOR adaptation training, consisting of self-generated (active) head impulses, which lasted ∼15 min. The head impulses consisted of rapid, horizontal head rotations with peak-amplitude 15 o , peak-velocity 150 o /s and peak-acceleration 3,000 o /s 2 . The VOR gain towards the adapting side increased after training from 0.92±0.18 to 1.11±0.22 (+22.7± 20.2 %) during active head impulses and from 0.91± 0.15 to 1.01±0.17 (+11.3±7.5 %) during passive head impulses. During active impulses, the VOR gain towards the non-adapting side also increased by ∼8 %, though this increase was ∼70 % less than to the adapting side. A similar increase did not occur during passive impulses. This study shows that unilateral vestibular adaptation is possible in humans with a normal VOR; unilateral incremental VOR adaptation may have a role in vestibular rehabilitation. The increase in passive VOR gain after active head impulse adaptation suggests that the training effect is robust.