This paper deals with the following three topics: (1) interfering noise (voice babble, single competing speaker) as the main problem of many hearing-impaired listeners, (2) the amplitude-frequency response of the hearing aid, and (3) the benefit of frequency-dependent compression. Research by the author and his coworkers has shown that: (1) persons with impaired hearing typically need 3 to 6 dB higher speech-to-noise ratios than do normal-hearing listeners--a technically very difficult problem to solve; (2) within a relatively ample range, the speech-reception threshold in noise is independent of the amplitude-frequency response; and (3) the small time constants of syllabic compression deteriorate the speech signal. Multichannel amplification (2-4 channels) with automatic gain control for each channel is recommended, optimally adjusted to keep the (variable) speech signal within the impaired ear's limited dynamic range as well as to preserve the intensity differences of successive speech phonemes.