The just-noticeable difference for the interaural time difference (ITD) of a sound generally reduces as its duration is increased [e.g., T. Houtgast and R. Plomp, J. Acoust. Soc. Am. 44, 807–812 (1968); E. Hafter and R. H. Dye, J. Acoust. Soc. Am. 73, 644–651 (1983)]. The rate of this temporal integration is less than would be expected from a simple multiple-looks application of signal detection theory, as though one part of the sound is given greater weighting. The present experiment measured the contributions of the start, middle, or end of a sound to the detectability of its ITD. The stimulus was a 32-pip train of 500-Hz, 10-ms pips, diotic apart from some (2–16) target pips which carried the ITD to be detected and which were placed either at the start, middle, or end of the train. Integration rates were determined from psychometric functions measured using four normal-hearing listeners. The results showed that the rate was least for targets placed at the start or end but larger for targets at the middle. The data can be described using a weighted multiple-looks approach, based on an integration function of approximately 100-ms time constant together with an emphasis of the start, and end, of the sound.