Three subjects with unilateral cochlear hearing loss and three subjects with bilateral cochlear hearing loss were tested in three experiments. In the first, their auditory filter shapes were measured for center frequencies of 700 and 2000 Hz, using the notched-noise method. The auditory filters were generally broader for the impaired than for the normal ears. In experiment 2, the threshold for detecting a 2000-Hz signal centered in a band of noise was measured as a function of the noise bandwidth for a Gaussian noise, and for that same noise multiplied (modulated) by a second noise low-pass filtered at 12.5 Hz. For the Gaussian noise, thresholds increased up to a certain noise bandwidth and then flattened off. This bandwidth was usually greater for the impaired than for the normal ears, consistent with the broader auditory filters of the impaired ears. For the modulated noise, thresholds tended to decrease when the noise bandwidth was increased beyond a certain value, indicating comodulation masking release (CMR). The decrease occurred at wider bandwidths for the impaired than for the normal ears. For the unilaterally impaired subjects, the amount of decrease was smaller for the impaired than for the normal ears when tested at equal SPL, but not when tested at equal SL. In experiment 3, the threshold for detecting a 700-Hz signal centered in a 20-Hz-wide band of noise (the on-frequency band, ONB) was measured in the presence of eight flanking bands (FBs) whose envelopes were either identical with that of the ONB (correlated condition) or were uncorrelated. CMR was defined as the difference in threshold between the correlated and uncorrelated conditions. The ONB and the FBs were presented either to the same ear (monaural condition) or to opposite ears (dichotic condition). CMRs tended to be greatest at high levels of the ONB and the FBs. CMRs in the monaural condition were smaller for hearing-impaired than for normal ears. However, at high levels, CMRs in the dichotic condition were similar for normal, bilaterally impaired, and unilaterally impaired subjects. In the latter case, CMRs were similar when the ONB was presented to the normal ear and to the impaired ear of each subject.(ABSTRACT TRUNCATED AT 400 WORDS)