A consonant identification test was carried out with 10 hearing-impaired listeners under various low-pass filter conditions. Subjects were also tested for cochlear dead regions with the TEN test. All subjects had moderate-to-severe high-frequency hearing losses. Consonant recognition was tested under conditions in which the speech signals were highly audible to subjects for frequencies up to the low-pass filter cut-off. Extensive dead regions were found for one subject with the TEN test. The remaining subjects may have had dead regions above 3 kHz, because of the severity of their hearing losses, but these could not be demonstrated with the TEN test. Average consonant scores for the subject group improved significantly (p<0.05) with increasing audibility of high-frequency components of the speech signal. There were no cases of speech perception being reduced with increasing bandwidth. Nine of the subjects showed improvements in scores with increasing audibility, whereas the remaining subject showed little change in scores. For this subject, speech perception results were consistent with the TEN test findings. In general, the results suggest that listeners with severe high-frequency losses are often able to make some use of high-frequency speech cues if these cues can be made audible.