Certain tiger moths emit high-frequency clicks to an attacking bat, causing it to break off its pursuit. The sounds may either orient the bat by providing it with information that it uses to make an attack decision (aposematism) or they may disorient the bat by interrupting the normal flow of echo information required to complete a successful capture (startle, jamming). At what point during a bat's attack does an arctiid emit its clicks? If the sounds are aposematic, the moth should emit them early in the attack echolocation sequence in order to allow the bat time to understand their meaning. If, however, the sounds disrupt the bat's echo-processing behaviour, one would expect them to be emitted later in the attack to maximize their confusion effects. To test this, we exposed dogbane tiger moths (Cycnia tenera) to a recording of the echolocation sequence emitted by a big brown bat (Eptesicus fuscus) as it attacked a stationary target. Our results demonstrate that, at normal echolocation intensities, C. tenera does not respond to approach calls but waits until the terminal phase of the attack before emitting its clicks. This timing is evident whether the moth is stationary or flying and is largely independent of the intensity of the echolocation calls. These results support the hypothesis of a jamming effect (e.g. 'phantom echoes') and suggest that, to determine experimentally the effects of arctiid clicks on bats, it is important that the bats be tested under conditions that simulate the natural context in which this defence operates.