Frequency analysis covers two separate listening tasks, one involving frequency discrimination, the other frequency selectivity. Discrimination refers to the ability to distinguish one frequency from another. Selectivity refers to the ability to hear one frequency in the presence of other frequencies. Selectivity is critical to the understanding of speech which comprises sounds containing many different frequencies. To understand speech easily, the listener must be able to analyze speech sounds into their component frequencies, especially formants. The hard-of-hearing person is probably less able to make that analysis, but we know surprisingly little about either discrimination or selectivity in hearing impairment. Existing evidence does suggest that both discrimination and selectivity are reduced in cochlear impairment so that such patients need a bigger frequency difference to discriminate between two tones and they have a wider critical band. A widened critical band would be expected to make it very difficult for the severely impaired person to understand speech under all listening conditions; it would make it difficult for the moderately impaired person to understand speech in a noisy background, unless the signal-to-noise ratio is improved as is possible by appropriate amplitude compression in hearing aids.