Usher syndrome has been historically categorized into one of three classical types based on the patient phenotype. However, the vestibular phenotype does not infallibly predict which Usher genes are mutated. Conversely, the Usher syndrome genotype is not sufficient to reliably predict vestibular function. Here we present a characterization of the vestibular phenotype of 90 patients with clinical presentation of Usher syndrome (59 females), aged 10.9 to 75.5 years, with genetic variants in eight Usher syndromic genes and expand the description of atypical Usher syndrome. We identified unexpected horizontal semicircular canal reactivity in response to caloric and rotational stimuli in 12.5% (3 of 24) and 41.7% (10 of 24), respectively, of our USH1 cohort. These findings are not consistent with the classical phenotypic definition of vestibular areflexia in USH1. Similarly, 17% (6 of 35) of our cohort with USH2A mutations had saccular dysfunction as evidenced by absent cervical vestibular evoked myogenic potentials in contradiction to the classical assumption of normal vestibular function. The surprising lack of consistent genotypic to vestibular phenotypic findings as well as no clear vestibular phenotypic patterns among atypical USH cases, indicate that even rigorous vestibular phenotyping data will not reliably differentiate the three USH types.