Recent neurophysiological findings, cited in previous publications imply that some vestibular commissural pathways may form positive feedback loops across the midline. It has already been shown theoretically that such feedback coupling of the vestibular nuclei could play an important role in the realization of the central integrator in the vestibuloocular reflex (VOR). In addition, it was found that known commissural plasticity during vestibular compensation, if placed at the level of such cross-midline loops, could reconcile findings after labyrinthine lesions. This paper examines theoretically the role such commissural feedback loops could play in the adaptation of the dynamics of the VOR in normal behaving animals. A simple static example is used to illustrate that changes in synaptic efficacy along cross-midline feedback loops could serve to adjust both balance and gain in vestibular reflexes. A bilateral model of the VOR and its interactions with vision is used to explore analytically the consequences of parametric changes along cerebellar and/or commissural pathways in three protocols: VOR in the dark, visual pursuit, and visual VOR suppression. Model predictions are systematically related to published findings after short- and long-term adaptation of the VOR. Conclusions arising from the theoretical results point to specific strategies that can be used in experiments on intact alert animals, in the further study of vestibular adaptation, and in the diagnosis of possible sites of plasticity. This should be relevant to arguments on cerebellar versus brain stem sites for vestibular adaptation, currently a highly controversial issue. For example, it is found that observations of responses in the adapted VOR in the dark are not sufficient to distinguish between a brain stem or cerebellar site for VOR plasticity. Also, the analysis shows that, in the model, changes in the VOR gain would often be associated with parallel changes in VOR dynamics; this has often been reported, but previously left unexplained. Model predictions of response changes in the adapted VOR, during VOR suppression, do provide a means of distinguishing between brain stem or cerebellar sites of plasticity; only the brain stem site, postulated here in the commissural loops, would produce cerebellar response changes compatible with the observations of Miles and Lisberger, during long-term adaptation of the VOR. A cerebellar site for VOR adaptation in the model would produce changes in cerebellar responses that would only be compatible with observations to date during rapid, or short-term (hours), modification of the VOR, as reported by Ito and his group.(ABSTRACT TRUNCATED AT 400 WORDS)