Associative lists created by the same means are remarkably different in their propensity to elicit false memories in the DRM (; ) paradigm. We confirmed this variability in Experiment 1 by constructing lists in the typical fashion but with words that were weakly associated to their critical words. Low levels of false recall occurred. In Experiment 2 these results were replicated at three presentation rates (.5, 1, and 3s per word). Also, slower presentation rates yielded lower false recall for both strong and weak lists. Experiment 3 showed that false recognition rates also varied across lists, as did subjective ratings accompanying false recognition. We interpret these findings as supporting an activation/monitoring framework. Lists vary in a principled way in their tendency to activate the critical item, and slowing the presentation rate permits greater accrual of item-specific information that makes monitoring of retrieval more accurate.