Caffeine's effects on recall of word lists were investigated using the Deese-Roediger-McDermott (DRM) paradigm. College students were administered either 200 mg of caffeine or a 250-mg lactose placebo; after 30 min., they were tested on recall using six word lists. Words of each list were semantically related to a single word (a "critical lure") that was not presented in the list. Participants administered caffeine recalled more list words and more critical lures than participants administered lactose. Recall of list words was negatively correlated with recall of critical lures. Caffeine appears to intensify the strength of connections among list words and critical lures, thereby enhancing both true and false memory.