In the flash-lag illusion, the location of a moving object aligned with a flashed stimulus is misperceived ahead of the flash. Two explanations have been offered. Nijhawan (Nature, 1994) proposes a predictive theory, which claims that the visual system corrects for transmission delays by using the motion of the object before the flash to extrapolate its position at the time of the flash. In contrast, Eagleman and Sejnowski (JoV07) have suggested a postdictive theory that argues that the motion of the object after the flash biases the perceived position of the object at the time of the flash. The postdictive theory correctly predicts that the flash-lag effect is not observed when the moving object disappears at the time of the flash. In order to account for this, the predictive theory requires an additional assumption, that a transient signal on the moving object at the time of the flash will override the extrapolation, yielding a veridical percept. To test this assumption, we presented moving items with three types of transients occurring at the same time as the flash. The moving object was changed from black to white, reduced from 30% to 4% contrast, or reduced from 4% to 0% contrast (i.e. it disappeared). In the first two conditions, the flash-lag effect was just as strong as in a control condition with no transient signals. When the moving item vanished, however, the illusion was entirely absent, even though the transient signal generated by the disappearance of this moving low contrast object was much less than that generated in the previous two conditions. Transient signals per se do not disrupt the flash-lag effect. Rather, the disappearance of the moving object, and the subsequent lack of motion signals, eliminates the effect. This finding contradicts the predictive account, but is consistent with a postdictive account.