Two experiments using a levels-of-processing paradigm were performed to demonstrate the existence and usefulness of a semantic component in chess knowledge. Experiment I compared forced-choice recognition of chess positions after a structural task (piece counting) as opposed to a semantic task (choosing a move). Recognition accuracy, confidence, and familiarity ratings all showed a facilitation effect in the semantic condition. By including an orienting task that did not encourage semantic processing but still allowed pattern-matching operations to occur (copying a board), Experiment II demonstrated that this task effect was a genuine enhancement of memory due to meaningful processing. One again, the processing of meaningful relations in the semantic task (positional evaluation) produced a higher level of recognition performance than did the more structural processing. These results suggest that aspects of meaning have some input into the processes that generate the memory representation of a chess position.