BACKGROUND and AIMS. During balance rehabilitation, automatic visual-vestibular compensations occur to reduce the patients’ visual dependence due to the vestibular inhibition suppression. This study presents the positive therapeutic effects of interactive protocols based on immersive virtual reality (VR) that restore the reciprocal visual-vestibular inhibition on a short period of two months, and the use of an identification procedure to characterize the relationship between visual stimulation features and the adaptive equilibrium parameters associated to the balance enhancement. METHODS. Included in an 8 weeks clinical trial, nine unilateral vestibular patients (42-80 y.o) were immersed into structured large-field visual virtual flows, whose the dynamic constraints and complexity (scroll, radial and rotation visual pattern, stimulation speed, and gaze anchoring) increase throughout the therapeutic session. The analysis focus on the gaze behavior, feet center-of-pressure trajectory (CoPT) and disequilibrium indicators. RESULTS. The results show that the balance perturbations and the associated compensatory postural strategies are intensified (increasing of the CoPT and ID values) related to the complexity of the virtual pattern and speed, but progressively decrease significantly throughout the sessions, for all the stimulation conditions and scenarios. The methodical repetition of the virtual scenarios, associated to the consistent practice through sessions, strengthened the adaptive balance improvement. Data indicates the balance recovery and a patients’ self-confidence due to the reduction of the visual dependence boosted by the visual-vestibular reverse compensation. CONCLUSION. At a clinical level, our results demonstrate the real and rapid effectiveness of the large-field visual immersion on condition of controlled and repetitive virtual scenario. Despite the lack of long-term effects (reported by patients on a six months period) requiring periodic booster sessions, new rehabilitation strategies including virtual visual immersions have to be preferred comparing to standardized rehabilitation protocols (uncontrolled optokinetic stimulation). At a neuro-functional level, the visual-vestibular-motor adaptation despite the persistence of vestibular deficits indicates the relative reweighting of the visual-vestibular sensory inputs for a functional reciprocal inhibition restoration. New questions on the neural basis of the visual-vestibular plasticity for balance control will be addressed. Finally, our study corroborate the proof of concept of controlled VR-rehabilitation, with experimental, practicable and clinical validation.