Listeners are presented with pulse-train stimulus pairs and asked to judge whether they can hear a difference between them. The interval between pulses is a random variable, identically and independently distributed for each stimulus of a pair. Two distributions are observed: one, nominally Gaussian, and the other, nominally the distribution of the amplitude of a sinusoidal wave whose phase is uniformly distributed. The principal experimental parameters are the mean interval between pulses and rms deviation or jitter about this interval. The stimuli of each pair are identical in other respects, pulse shape (50-μsec pulse width), loudness (30 or 35 dB SL), but differ in polarity pattern. Two pattern combinations are observed. For the range of mean pulse intervals investigated, 5–15 msec, the stimuli of each pair are generally discriminable when unjittered. However, the results indicate that, when jittered in amounts greater than 1 or 1.5 msec, the stimuli may be rendered nondiscriminable. This critical amount of jitter coincides with a flattening of the power-density spectra of the stimuli for frequencies greater than 250 or 300 cps. On the basis of this result and the results of other investigations, it is hypothesized that the correlates of discrimination for unjittered or lightly jittered stimuli are distinct neural volley patterns associated with basilar membrane activity in the 300- to 1000-cps region.