The brain's ability to bind incoming auditory and visual stimuli depends critically on the temporal structure of this information. Specifically, there exists a temporal window of audiovisual integration within which stimuli are highly likely to be perceived as part of the same environmental event. Several studies have described the temporal bounds of this window, but few have investigated its malleability. Recently, our laboratory has demonstrated that a perceptual training paradigm is capable of eliciting a 40% narrowing in the width of this window that is stable for at least 1 week after cessation of training. In the current study, we sought to reveal the neural substrates of these changes. Eleven human subjects completed an audiovisual simultaneity judgment training paradigm, immediately before and after which they performed the same task during an event-related 3T fMRI session. The posterior superior temporal sulcus (pSTS) and areas of auditory and visual cortex exhibited robust BOLD decreases following training, and resting state and effective connectivity analyses revealed significant increases in coupling among these cortices after training. These results provide the first evidence of the neural correlates underlying changes in multisensory temporal binding likely representing the substrate for a multisensory temporal binding window.