The timing of visual information pick-up for visual anticipation was examined by comparing the capability of multiple skill groups, expert and near-expert karate athletes and novices, to block attacks using an in situ temporal occlusion paradigm. Participants stood facing a karate opponent and then attempted to block attacks (kicks and punches), whilst their vision of attacks was occluded: (a) prior to onset of opponent motion (O1), (b) after preparatory head movement (O2), and (c) after initiation of the attacking motion (O3). A no occlusion control condition provided complete vision of attacks (O4). Results revealed that expert anticipation was not significantly different to that of near-experts at O1, but was significantly different to the other group across O2-O4. Expert anticipation, however, was significantly above chance across all occlusion conditions, but near-experts performed above chance at O3 and O4, whilst novices were better than chance at O4. Unexpectedly, unique evidence was found that expert anticipation could be differentiated from near-expert anticipation in the earliest occlusion condition, where it was found that only experts were capable of using visual information from a static opponent to anticipate and block attacks above chance. The findings further understanding of expert visual anticipation to guide motor skills beyond existing expert-novice comparisons.