An important challenge for auditory neuroscience is explaining the deficits in sound perception in persons with sensorineural hearing impairment. The most common example of this impairment is the, usually high frequency, hearing loss of old age. The most obvious deficit is a loss in threshold sensitivity which renders sounds inaudible. However, even when audibility is corrected with amplification, as in a hearing aid, normal hearing may not be restored. The reasons for the remaining deficits are not fully understood. In this talk, the nature of the perceptual deficits caused by sensorineural hearing loss will be reviewed, along with work on the relationships between the cochlear lesions that are present and the pathologies in auditory neural responses to sound. Prominent among these is a decrease in the sharpness of tuning of auditory nerve fibers, i.e. a decrease in frequency selectivity. By studying the responses to speech-like sounds in ears with sensorineural loss, it is possible to demonstrate corresponding severe deficits in the representation of the sounds' spectra. For example, the analysis of sound by frequency, analagous to a Fourier decomposition, which occurs in a normal cochlea, is partially or fully lost in the presence of sensorineural loss. Ordinarily, the different frequency components which make up a speech sound are separated by frequency along the basilar membrane. This means that the formant frequencies of speech (corresponding to the resonant frequencies of the vocal tract), induce responses in different populations of auditory nerve fibers, according to frequency. In the presence of sensorineural damage, responses to the first formant are observed in essentially all auditory nerve fibers and there is no place in the cochlea where responses exclusively to the second and third formants, which are essential to good speech understanding, are found. The implications of these results for speech perception and possible spectral modifications of the speech stimulus to overcome these deficits primarily by Roger Miller, with the collaboration of the author, and was supported by NIH grants.