The perceptual segregation of simultaneous speech is known to benefit from differences in pitch or loudness between the speech sounds. It is likely that different cues operate on different time courses. A series of experiments examined subjects' ability to identify both members of simultaneously presented steady‐state vowel pairs. The duration of each vowel pair varied from 1 to 8 cycles. A “dominant” vowel was identified with near 100% accuracy even in the single‐cycle vowel pair condition (one pitch period), whereas the recognition of the “nondominant” vowel showed a slow improvement up to the 8‐cycle condition. Asynchrony between the pitch periods of the vowels improved performance when they had a low fundamental frequency (80 Hz), but about 3 cycles of the vowels were necessary for this advantage. The effects of changing the amplitude of one vowel relative to the other and the effects of differences in frequency modulation were also studied. The results support previous research in showing that phonetic decisions “await” organizing mechanisms that operate in the auditory periphery. [Work supported by Esprit BRA 3207.]