The representation of item and associative information in episodic memory was investigated using cued recall and single item recognition. In the first four experiments, participants studied two lists constructed such that some items presented in a pair during List 1 were rearranged to create new pairs in List 2 and were accompanied by pairs exclusive to List 2. List 1 was composed of either word-face pairs (Experiment 1 and 3) or word-word and face-face pairs (Experiment 2 and 4). Participants were tested for their memory of the second list containing only word-face pairs. When the test required associative information (i.e., cued recall), interference was specific to pair-type. Specifically, repeating items in the same pair-type across lists led to a greater number of correct and incorrect responses but repeating items in different pair-types did not change performance. When the test did not require associative information (i.e., single item recognition), interference was not specific to pair-type; hits and false alarms were higher for items presented on List 1. In the final experiment, the number of studied pairs and single items was independently manipulated and a pair-specific list length effect was observed. That is, performance in cued recall was modulated by the number of studied pairs, but not the number of studied individual items. Together, these data suggest that information specific to the pair-type (i.e., associative information) is stored and strategically utilized during memory search. (Contains 5 figures and 4 tables.)