We examined whether the movement involved in a pointing gesture, depicted using point-light displays, is sufficient to cue attention in typically developing children (TD) and children with autism spectrum disorder (ASD) (aged 8–11 years). Using a Posner-type paradigm, a centrally located display indicated the location of a forthcoming target on 80 % of trials and the opposite location on 20 % of trials. TD children, but not children with ASD, were faster to identify a validly cued target than an invalidly cued target. A scrambled version of the point-light pointing gesture, retaining individual dot speed and direction of movement but not the configuration, produced no validity effect in either group. A video of a pointing gesture produced validity effects in both groups.