Falls, caused by dizziness and balance deficits, and their consequences in terms of the resultant functional limitations and of the associated costs for the involved actors as well as for the healthcare system are a serious problem in modern societies. Vestibular rehabilitation therapy (VRT) is proposed as a personalized intervention both for ameliorating symptoms and for improving balance. The outcomes of the therapy are not the same in all patients and the reasons are explored in this study with data mining methods applied in two retrospective datasets. According to our findings, patients with unstable lesions, concurrent psychological problems and with headaches are poorer candidates for VRT. Patients with stable vestibular deficits, of recent onset and relatively intact psychological, acoustic, visual and proprioceptive systems are better candidates for VRT.