Many researchers have proposed that when an individual observes the actions of another individual, the observer simulates the action using many of the same neural areas that are involved in action production. The present study was designed to test this simulation hypothesis by comparing the perception of multisensory stimuli during both the execution and observation of an aiming action. The present work used the fusion illusion - an audio-visual illusion in which two visual stimuli presented with one auditory stimulus are erroneously perceived as being one visual stimulus. Previous research has shown that, during action execution, susceptibly to this illusion is reduced early in the execution of the movement when visual information may be more highly weighted than other sensory information. We sought to determine whether or not a non-acting observer of an action showed a similar reduction in susceptibility to the fusion illusion. Participants fixated a target and either executed or observed a manual aiming movement to that target. Audiovisual stimuli were presented at 0, 100, or 200 ms relative to movement onset and participants reported the number of perceived flashes after the movement was completed. Analysis of perceived flashes revealed that participants were less susceptible to the fusion illusion when the stimuli were presented early (100 ms) relative to later in the movement (200 ms). Critically, this pattern emerged in both execution and observation tasks. These findings support the hypothesis that observers simulate the performance of the actor and experience comparable real-time alterations in multisensory processing.