Purpose: Previously, we reported that synchrony perception in stochastic displays is tuned for orientation and spatial frequency but not for spatial separation (VSS 2001). Here, we used a masking paradigm to reveal the temporal properties of mechanisms mediating synchrony perception. Method: Observers discriminated between pairs of synchronous and non-synchronous Gabor elements in a 2AFC task. Spatially, Gabors had a 3 deg horizontal separation and vertical 2.0 cpd carriers with random initial phases. Temporally, each Gabor was driven by a 533 ms waveform that modulated spatial phase over time. This waveform consisted of the linear sum of white Gaussian noise derived from two sources: a synchrony source S common to both elements and an independent masking source M for each element. We separately filtered the S and M components into one of seven 1.2-octave log-Gaussian bands (3.5 to 30 Hz) and measured synchrony thresholds for all 49 pairs of S and M bands. Defining synchrony as the ratio between the variances of S and S+M, we manipulated synchrony and estimated 75%-correct thresholds. Results: When peak frequencies of the S and M bands coincided, synchrony thresholds remained constant over all frequencies tested. For high-frequency S bands, thresholds decreased as a function of frequency separation between S and M bands. For low-frequency S bands, however, we found little evidence of threshold reduction as we increased frequency separation between S and M bands. Conclusions: Same-band masking revealed synchrony detection in signals is possible over temporal frequencies spanning a range of at least 3.5 to 30 Hz. Cross-band masking revealed synchrony detection relies on bandpass channels at high frequencies but on broadband channels at lower frequencies. These data, combined with data on the spatial properties of synchrony mechanisms, provide evidence for spatiotemporal association fields that mediate the detection of synchronous events in stochastic displays.