The definition of ototoxicity in most clinical studies of aminoglycoside antibiotics is an increase in pure-tone threshold from a baseline audiogram greater than or equal to 15 dB at two or more frequencies, or greater than or equal to 20 dB at one or more frequencies. In this study, test-retest auditory threshold differences of this magnitude were found in a group of 20 normal volunteers who were not taking any known ototoxic drugs. Depending on which of the two criteria for ototoxicity are used, these data represent a 20% or 33% incidence of ototoxicity. We believe that many of the audiometric changes reported to represent aminoglycoside antibiotic ototoxicity may actually represent the normal test-retest variability of pure-tone audiometry. If this is true, the reported incidence of hearing loss due to aminoglycoside antibiotics may be exaggerated.