When two very short time intervals are presented serially by sound markers (in such a way that they share a common marker) the subject's duration judgments of the second time interval can be affected by the duration of the first interval. Such a conspicuous effect has not been reported in the literature. Standard empty time intervals of 120, 240, 480, and 720 msec were preceded by a neighboring empty time interval of various physical durations, and subjects adjusted a comparison empty time interval to the same subjective duration as these standards. We found clear underestimations of the standard duration when its physical duration was 120 msec. For example, when the preceding duration was 45 msec, the relative underestimation was about 40%. Because such a stable and remarkable underestimation appeared in a very simple situation, this phenomenon may be called a new illusion. Such an illusion did not appear when the time interval to be judged was succeeded by another time interval. At present we cannot explain the illusion, but in the general discussion we attempt to relate it to some findings in rhythm perception.