The present study investigated the relative importance of temporal and spectral cues in voice gender discrimination and vowel recognition by normal-hearing subjects listening to an acoustic simulation of cochlear implant speech processing and by cochlear implant users. In the simulation, the number of speech processing channels ranged from 4 to 32, thereby varying the spectral resolution; the cutoff frequencies of the channels' envelope filters ranged from 20 to 320 Hz, thereby manipulating the available temporal cues. For normal-hearing subjects, results showed that both voice gender discrimination and vowel recognition scores improved as the number of spectral channels was increased. When only 4 spectral channels were available, voice gender discrimination significantly improved as the envelope filter cutoff frequency was increased from 20 to 320 Hz. For all spectral conditions, increasing the amount of temporal information had no significant effect on vowel recognition. Both voice gender discrimination and vowel recognition scores were highly variable among implant users. The performance of cochlear implant listeners was similar to that of normal-hearing subjects listening to comparable speech processing (4-8 spectral channels). The results suggest that both spectral and temporal cues contribute to voice gender discrimination and that temporal cues are especially important for cochlear implant users to identify the voice gender when there is reduced spectral resolution.