The reliability of a closed‐response set nonsense syllable test [S. B. Resnick, J. R. Dubno, S. Hoffnung, and H. Levitt, J. Acoust. Soc. Am.58,114 (1975)] was determined on a group of 38 listeners with mild to moderate sensorineural hearing loss. Eight randomizations of the 91‐item test (four trials on each of two days) were presented monaurally, under earphones, at 90 dB SPL with a cafeteria background noise set at a +20 dB speech‐to‐noise ratio. Performance under these conditions ranged from 21.4% to 91.2%, reflecting the wide range of syllable‐recognition ability of these subjects. Reliability of the eight measurements was determined by analysis of variance and analysis of covariance structure (parallel‐test modelling) for the entire test and each of 11 subtests. Overall and individual subject results failed to show any systematic differences in scores over eight trials. Likewise, no significant differences were found in performance on individual syllables, nor were changes in the relative occurrence of specific syllable confusions noted. This test is highly reliable when evaluating hearing‐impaired subjects, and thus is appropriate for use in investigations where identical items are administered under multiple experimental conditions.