We examined differences in the methods in which collegiate kendo players (mid-level skill group, n=7; high-level skill group, n=8) utilized contextual information in response selection. The subjects participated in four matches each. Immediately after the end of each match, they observed a video of the match and replied to a semi-structured interview. We used the replies to analyze the methods in which the players utilized contextual information, including what information was collected and how it was collected, and what information was utilized during the matches. In response selection, the high-level skill group had significantly more methods of information utilization based on information collected during the matches. They collected opponents' information efficiently through various events in various scenes, and then they effectively utilized this information. On the other hand, the mid-level skill group could not efficiently collect opponents' information and had a higher frequency of response selection without utilizing collected information. Furthermore, our results showed to score points (yuko-datotsu), it was effective to collect opponents' information “contextually” and to utilize this information in response selection. We therefore believe that acquisition of the following abilities should be priorities among kendo players; increased perceptual-expertise and knowledge of how to collect information on an opponent; development of processing functions to bring about effective information storage and retrieval; and acquisition of various motor skills to facilitate information utilization.