Numerous studies have demonstrated that adaptation to fixed levels of audiovisual asynchrony recalibrates perceived simultaneity (Fujisaki et al., 2004; Heron, et al., 2007). However, the mechanism(s) underlying this phenomenon remain poorly understood. In particular, it is not known whether temporal recalibration is a mandatory, stimulus-driven process, or whether it is subject to cognitive control. Here, we address this issue by manipulating observers’ attentional state during adaptation to audiovisual asynchrony. Within a given experimental session, observers were instructed to detect one of three different oddball stimuli (i) a change in the contrast polarity of the fixation cross: ‘attend fixation condition’, (ii) a change in the size/position of the auditory and visual stimuli: ‘attend stimuli’ condition, or (iii) a change in the temporal order of the adapting stimulus pairs: ‘attend temporal order’ condition. Critically, the sequence of adapting stimuli itself remained unchanged across conditions. Whilst some degree of adaptive temporal recalibration was found in all conditions, attending to temporal order proved by far the most effective manipulator of adaptive magnitude. Thus, in keeping with other adaptation phenomena such as the motion aftereffect (Mather, 1998), it appears that asynchrony adaptation can be substantially modulated by goal driven, top-down processes.