Presentation of test signals consisting of sounds with rippled spectra allowed measurements of the frequency-resolving ability (FRA) of hearing to be performed in humans without using frequency-dependent masking techniques. This allowed studies of changes in FRA in the presence of noise interference. In conditions of diotic presentation (to both ears in parallel) of the test signal and noise, FRA decreased significantly if the interference frequency was lower or equal to the test signal frequency. The relationship between this effect and the sound intensity and noise:signal ratio varied for low-frequency noise and for noise at the same frequency as the test signal, which indicates that these two types of interference have different mechanisms. However, in both cases, noise of sufficient intensity led to a complete inability to discriminate the fine spectral structure of the test signal. In dichotic presentation (test signal to one ear, noise to the other), noise had virtually no effect on FRA over a wide range of test signal and noise frequencies and noise:signal ratios. Thus, there was essentially complete dichotic release of FRA from the effects noise, which has potential to be used in constructing prosthetic hearing devices.