Objectives: Determine the prevalence of otolith dysfunction in patients who have failed treatment for benign paroxysmal positional vertigo (BPPV). Methods: Case-control study of patients with BPPV who failed standard treatment and subsequently underwent vestibular testing and supervised vestibular rehabilitation. Interventions included videonystagmography, rotary chair, subjective visual vertical test (SVV), and cervical vestibular evoked myogenic potentials (VEMP). We compared the prevalence of abnormalities in tests of otolith dysfunction (SVV and VEMP) in patients who responded or failed vestibular rehabilitation. Results: In 2012, 46 of 251 patients with BPPV failed initial treatment with canalith repositioning maneuvers performed in the office; of these 46 patients, 28 patients had posterior semicircular canal BPPV and 18 cases had atypical presentations (multicanal BPPV, anterior canal BPPV, or signs of uncompensated unilateral peripheral vestibulopathy). Vestibular testing followed by customized vestibular rehabilitation was completed in 40 patients: 21 patients had resolution of their positional vertigo and nystagmus, and 19 patients did not respond to therapy. Abnormal otolith tests (SVV and/or VEMP) were more common in patients who failed therapy (Pearson Chi square, P = 0.002). Conversely, abnormalities of caloric testing did not predict response to therapy. Conclusions: Abnormalities of utricle and saccular function are prevalent in patients with refractory BPPV. This fact is important to take into account when designing rehabilitation strategies for BPPV patients who fail canalith repositioning maneuvers.