The perception of an intervocalic b can be changed to p by lengthening the silenced closure interval. The slower the speaking rate, the longer the closure interval needed, roughly in proportion to the duration of the preceding syllable [Port, J. Acoust. Sec. Am. Suppl. 1 63, S20 (1978)]. To see whether the structure of that preceding syllabic also affects the voicing boundary, slowly spoken /dab/ syllables were shortened to match the durations of medium rate and fast rate /dab/ syllables by deleting pitch pulses from the steady-state region of the vowel. A variable silent interval and a constant /bi/ were appended to each of the /dab/ syllables, making /dabi/ to /dapi/ continua. The voicing boundary was at a nearly identical ratio of closure duration to preceding syllable duration for unmodified syllables at all speaking rates, but it was at a lower ratio (proportionally less silence was needed) for the originally slow syllables in which the vowel had been shortened. In fact, it was found using synthetic speech that longer syllables with proportionally short steady-state sections needed less silence than shorter syllables with proportionally long steady-state sections. Thus, the perceptual voicing boundary is sensitive to the dynamic structure of the preceding syllable and not simply its duration. [Research supported by NICHD.]