Two clinical word discrimination tests, the Northwestern University Auditory Test No. 6 (NU-6) and the Modified Rhyme Test (MRT), were examined for internal consistency and for split-half reliability under varying primary-to-secondary ratios (P:S) in normal hearers. Although the statistical analysis indicated that half-lists could be used reliably for both discrimination tests at most P:S conditions, the wide variability found among individual listeners does not support the clinical practice of half-list testing in noise-at least not for normal-hearing listeners.