Gardiner and Java (1990) report remember/know/new data that heavily favor the dual process account. In their Experiment 2, participants (N=20) studied 15 words and 15 non-words and, following a 24 hour delay, their memory was tested with an old/ new recognition procedure (30 old, 30 new items). Crucially, for items classified as old Gardiner and Java required participants to make an additional classification. Participants were instructed to respond “remember” when they could recall specific contextual details, or respond “know” when they simply felt like they had encountered the item. Their data (their Figure 2) demonstrated a clear cross over interaction, such that words elicited far more “remember” responses, whereas non-words elicited more “know” responses. Importantly, the frequency of “remember” and “know” responses did not differ between unstudied (new) words and non-words. The cross over interaction for old items in conjunction with a lack of effect for new items is highly constraining for theories of recognition and strongly favor a dual process model. Given their potential theoretical significance, the present study aims to replicate the findings of Gardiner and Java (1990).